Results: https://g.co/gemini/share/f4234cab8a2f

Analysis Tool: Gemini

Results: https://g.co/gemini/share/4f996dec9973

Analysis Tool: Gemini

https://g.co/gemini/share/756b404a5d02

My Concerns with DeepSeek's Terms of Use

1. Who Owns Your Ideas (Inputs and Outputs)?

The Big Idea: When you use DeepSeek, you'll be giving it "Inputs" (like your questions or information) and it will give you "Outputs" (the answers or content it creates). It's super important to know who owns these ideas.

My Concern: Section 4.2 says:

• You keep your rights to your Inputs. That's good! If you type in your secret recipe, it's still your secret recipe.

• DeepSeek gives you the rights to the Outputs it creates for you. This sounds great because it means you own the cool stuff the AI makes for you! You can use it for "personal use, academic research, derivative product development, training other models (such as model distillation), etc."

Why I'm Worried: Even though they say they assign you the rights to the Outputs, Section 4.3 adds a twist. It says, "we may, to a minimal extent, use Inputs and Outputs to provide, maintain, operate, develop or improve the Services or the underlying technologies supporting the Services." They mention "secure encryption technology processing, strict de-identification rendering, and irreversibility to identify specific individuals." This means they'll try to hide who you are.

However, the main concern for me, especially with AI, is that they might use your Inputs and the AI's Outputs to make their AI smarter, even if they try to make it so your personal information isn't directly attached. While they give you an "opt out" option by turning off "Improve the model for everyone," many people might not notice this or understand its full implications. It's a bit like if you teach a robot how to bake, and then the robot uses what you taught it to become a super baker for everyone else, even if they don't say you taught it. The terms allow them to use your interactions to train and improve their own core AI models, which are a big part of their intellectual property, without clearly stating how that benefits you.

2. What if the AI Makes Mistakes or Bad Stuff?

The Big Idea: DeepSeek is powered by AI, and AI can make mistakes. The terms say you're responsible for what you put in and what you do with what the AI gives you.

My Concern: Section 4.4 and 4.6 are very clear:

• "All Outputs provided by this service are generated by an artificial intelligence model and may contain errors or omissions, for your reference only. You should not treat the Outputs as professional advice."

• "Any judgment you make based on the Outputs or subsequent related actions you take will result in consequences and responsibilities borne by you alone..."

• "DeepSeek can provide incomplete, inaccurate, or offending output."

This means if you ask the AI for medical advice, legal help, or anything serious, and it gives you wrong information that causes a problem, it's on you. DeepSeek is saying, "We told you it might be wrong, so it's your fault if you trust it completely." This is a strong statement that puts a lot of responsibility on you, the user, even if the AI system itself makes a significant error. From an ethics standpoint, while users should be careful, companies developing powerful AI also have a responsibility to minimize harm.

3. Deleting Your Account Doesn't Mean All Your Data is Gone

The Big Idea: When you delete your account, you'd think all your information would be erased.

My Concern: Section 2.5 says that even after you delete your account, DeepSeek has the right to "Retain certain data of the user as required by laws and regulations." This is pretty standard for legal reasons. However, they also say they can "Exercise the rights stipulated in these Terms for any illegal or violating behavior committed by the user during the use of the Services before deletion."

This means that even if you delete your account, any data they collected from your "illegal or violating behavior" could still be used against you or retained for their own purposes, which could include analyzing past interactions to improve their systems (even if that improvement is focused on detecting misuse). It's a reminder that once information is shared online, it can be hard to make it completely disappear.

4. Limited Guarantees from DeepSeek

The Big Idea: Companies usually offer certain promises about their products. DeepSeek is being very careful about what they promise.

My Concern: Section 7.3 says that the services are provided "AS IS" and "AS AVAILABLE." This means they're not promising that the service will always work perfectly, never have errors, or always be available. They specifically say they don't guarantee:

• "THAT YOUR USE OF THE SERVICES WILL MEET YOUR REQUIREMENTS"

• "THAT YOUR USE OF THE SERVICES OR ANY PART THEREOF WILL BE UNINTERRUPTED, TIMELY, SECURE OR FREE FROM ERROR"

• "THAT ANY OUTPUT OR OTHER INFORMATION OBTAINED BY YOU AS A RESULT OF YOUR USE OF THE SERVICES WILL BE ACCURATE, UP-TO-DATE, RELIABLE, NON-INFRINGING OR SECURE"

This is a very broad "no promises" statement. While common in software terms, it means you're largely on your own if the service doesn't perform as you expect or if the information it provides isn't reliable. For AI, where accuracy is a big deal, this is a significant point.

What I'd Like to See More Clearly Addressed:

I'd really like DeepSeek to be more specific about:

• How exactly are my Inputs and Outputs used to "improve the Services or the underlying technologies" if I don't opt out? Specifically, can my data, even if de-identified, contribute to the core training of future AI models that DeepSeek then sells or licenses to others?

• What are the specific measures they take to ensure "strict de-identification rendering, and irreversibility to identify specific individuals" when using Inputs and Outputs for improvement?

• What is DeepSeek's responsibility if their AI generates content that directly causes harm due to a factual error or bias, especially if the user had no way of easily verifying the information? While they disclaim responsibility, there's an ethical line for AI developers in minimizing the creation of harmful outputs.

These terms highlight the ongoing challenges in balancing user rights with the need for AI companies to improve their products using user data.

Do these concerns make sense, and do they help you understand the potential issues in these terms?

Based on my review of Anthropic's Terms of Service and related legal analysis, let me explain the key intellectual property concerns from a lawyer's perspective in simple terms:

Main Concerns About Who Owns What You Create

The Big Problem: Nobody Knows for Sure Anthropic says they give you "all of our right, title, and interest—if any—in Outputs," but notice those important words "if any." This is like saying "I'll give you my piece of the cake, if I actually have any cake to give." They're not sure they own anything to begin with.

Why This Matters When you use Claude to write something, the law says only humans can own copyrights. It's like how only people can win spelling bees - robots can't. So if Claude writes a story by itself, nobody might own that story at all. It could be "free for everyone" (what lawyers call "public domain").

Your Responsibilities Are Heavy

You Must Own Everything You Put In The terms say "you represent and warrant that you have all rights, licenses, and permissions that are necessary for us to process the Prompts," which means if you feed Claude someone else's copyrighted work, you could get in big trouble. It's like promising you own every ingredient before you bake a cake with someone else's recipe.

You Can't Trust Everything Claude Says Anthropic warns that "Outputs may not always be accurate and may contain material inaccuracies even if they appear accurate because of their level of detail or specificity." This means you have to fact-check everything, like having to verify that your calculator didn't make mistakes.

Special Rules for Important Stuff

Some Uses Need Human Experts For important areas like medicine, law, or news, a "qualified professional in that field must review the content or decision prior to dissemination or finalization." Think of it like having a grown-up check your math homework before you turn it in.

What This Means for You

The Good News: You can use Claude's help and potentially own what you create together.

The Worrying News:

• You might not own anything Claude creates by itself

• You're responsible if something goes wrong

• You need human experts for important work

• The laws are still being figured out

My Advice: Use Claude like a very smart research assistant, but always add your own creative work, check everything carefully, and be honest about getting AI help. Think of it as collaboration, not replacement for human creativity.

This legal area is changing fast, like trying to make rules for a new sport while people are still playing it. Stay informed and be careful, especially for important or valuable work.

Retry

Claude can make mistakes.
Please double-check cited sources.

The Big Problems I See

1. Your Creative Work Gets Used Without Fair Payment When you chat with Gemini, Google uses this data to provide, improve, and develop Google products and services and machine-learning technologies, including Google's enterprise products such as Google Cloud. This means if you write a poem, story, or come up with a clever business idea in your chat, Google can use that to make their AI smarter and sell better products. You don't get paid for this, even though your creativity is helping them make money.

2. Human Strangers Read Your Private Conversations To help with quality and improve our products, human reviewers (including third parties) read, annotate, and process your Gemini Apps conversations. Imagine writing in what you think is a private diary, but strangers can read it and take notes. Even worse, conversations that have been reviewed or annotated by human reviewers are not deleted when you delete your Gemini Apps activity because they are kept separately and are not connected to your Google Account. So even if you try to delete your private thoughts, copies still exist for up to three years.

3. You Can't Really Turn Off Data Collection Even when Gemini Apps Activity is off, your conversations will be saved with your account for up to 72 hours. It's like having a recorder that you can't fully turn off - it always keeps the last few days of everything you say.

4. Your Personal Information Gets Shared Around When you connect other apps to Gemini, Gemini may share the following data with apps that are connected: Information from your conversation and your device, Preferences such as language, Location information. Your personal details get passed around to different companies, each with their own rules about what they can do with your information.

5. The "Legal Basis" Loophole Google claims they have legitimate interests to use your data, which is like saying "we think we have good enough reasons, so we don't need to ask permission." This gives them very broad power to use your conversations and ideas.

Why This Matters for Regular People

Think of it this way: if you had a smart friend who remembered everything you told them, then sold that information to companies and let strangers read your conversations, you'd probably stop talking to that friend. But with AI, we're often not told clearly how our words and ideas are being used to make these companies richer.

The biggest ethical problem is that people think they're just chatting with a helpful computer, but they're actually giving away their thoughts, creativity, and personal information to be used in ways that primarily benefit the company, not them.

Tool used: Google Gemini

Concerns About Your Privacy Policy

This policy is a very long and detailed document, and that's the first thing that makes me a little worried. Imagine a really big book with tiny words – it's hard for anyone, even grown-ups, to read and truly understand everything. Here are my main concerns:

1. Too Much Sharing, Too Many Friends

Your policy talks a lot about sharing your information with many "Third Parties." Think of it like this: if you tell one friend a secret, it's pretty safe. But if you tell that secret to 100 friends, and those friends can tell their friends, it's not a secret anymore!

• Who gets your information? The policy lists many different types of "friends" (like "Advertising and Analytics Partners," "Social Media and External Platforms," "Business Collaborators," and many more). It's hard to keep track of everyone who might get your information.

• What kind of information do they get? For things like your Contact Information and Transaction Information, the policy says it can be "Sold/Shared for Targeted Advertising" with "Third Party Advertising and Analytics Partners" and "Social Media and Other Third Party Platforms." This means your name, email, what you buy, and even what you just looked at might be used to show you ads everywhere you go online.

• Sensitive Information: The policy says you collect "Data Revealing Racial Origin" (Section 1.K). This is super sensitive information, like telling someone about your family background. While the policy doesn't list who this specific data is "Sold/Shared To" for advertising, it's still a big deal that it's collected at all, and it's not clear enough why it's needed or who exactly gets to see it. It's listed right there in Section 1, which is a bit unsettling given its sensitive nature.

2. "Other Uses" and "Other Service Providers" - What Do They Mean?

In many places, the policy says your information can be used for "Other Uses" or shared with "Other service providers who use your data solely to provide services to us or you."

• This is like saying, "We might do other things with your information, but we're not going to tell you exactly what those things are right now." It leaves a lot of open questions. We don't know who these "other" service providers are or what they're doing with your data. It's like inviting someone new to your house but not knowing their name or why they're there.

3. Using AI - Not Enough Details

Since my special interest is AI ethics, I looked for how AI might be used with your information. The policy mentions "Internal Operations" like "analyse trends," "troubleshoot issues," and "assess the effectiveness of our marketing efforts." While not directly stated, these are common areas where Artificial Intelligence (AI) or machine learning might be used to process huge amounts of data.

• AI's Power: AI can be very powerful, but it's important to know how it's being used, especially with your personal information. For example, if AI is used to "personalize" what you see, it could be guessing things about you based on your data. This is often called "profiling."

• Racial Origin and AI: When you collect "Data Revealing Racial Origin," and then combine it with "Internal Operations" that use AI, it raises questions. Could AI use this sensitive information to make decisions about you, like what job ads you see or what offers you get? The policy doesn't explain how AI might use this sensitive data, which is a big concern for ethical AI use. There's a risk of algorithmic bias, where the AI might unintentionally treat different groups of people differently because of the data it's trained on.

4. "De-identified or Pseudonymized Data" - Still a Risk?

The policy mentions that they might "de-identify or pseudonymize your personal data to render it non-personal." This means they try to remove your name or other direct ways to know it's you.

• Can it be un-de-identified? While they say they "will hand such de-identified or pseudonymized data as non-personal data, in line with the limits set by applicable laws" and "manage and utilize de-identified data without attempting to re-identify it," it's becoming easier with advanced technology and AI to put pieces of information back together and figure out who someone is, even if it's been "de-identified." The policy states they won't try to re-identify it "except where allowed by law, such as to verify the effectiveness of our de-identification processes." This "except where allowed by law" part leaves a bit of an opening.

In Simple Terms:

This privacy policy is like a big puzzle with some missing pieces and a lot of tiny instructions. It gives Solvely a lot of room to share your information with many different companies for many different reasons, some of which are not perfectly clear. When you add in the collection of very sensitive information like your "Racial Origin" and the possibility of AI being used, it makes me worry about how your personal data might be used in ways you don't expect or fully understand.

It's important that companies are very clear and specific about what they do with your information, especially in today's world where AI can learn so much from our data.

Do you have any questions about specific parts of the policy, or would you like me to explain any of these concerns in more detail?

My Main Concerns

1. What happens to your school work and ideas?

   • When you use Brisk Teaching, you put in your lesson plans, your ideas, and even some information about your students' work, like what they're learning or their grades. Brisk Teaching uses "AI helpers" (smart computer programs) to make its tools work. My big question is, does Brisk Teaching promise that your ideas and your students' work will only be used to help you and your students? Or could your information accidentally be used to teach the AI helpers to be smarter for everyone else who uses Brisk Teaching? It's like putting your special recipe into a unique mixer – you want to be sure no one else learns your secret recipe just from seeing you use your mixer. The notice says they share data with "AI service providers," which are the companies that make these smart AI helpers. We need to be very clear if your information helps their AI helpers learn generally, or just helps you specifically.

2. Keeping student information super private and fair.

   • Brisk Teaching uses information about students, which is very private. When these AI helpers look at private student information, we have to be extremely careful. My worry is that the AI helpers are used in a way that is always fair to every student. We don't want the AI to accidentally learn something unfair or make biased suggestions because of the information it sees. Also, even though it says they don't sell data, when they share student data with AI helpers, we need to be very sure that this private student information isn't used to teach the AI helpers in a way that could ever cause problems for a student later on.

3. Who owns the new things the AI helps you create?

   • If Brisk Teaching's AI tools help you write a new lesson plan, grade a paper, or create something new, who gets to say they "own" that new thing? Is it truly yours because you used your ideas and talent, or does Brisk Teaching have some claim because their AI helped you? The privacy notice doesn't clearly explain this. It's like if you use a very fancy art robot to draw a picture – who truly owns that finished drawing, you or the company that made the robot? It's important to know so there are no surprises later.

Overall, while the privacy notice mentions good things like not selling your data, it could be clearer about how your unique information, especially student data, is handled when it touches those smart AI helpers.

You can read the full privacy notice here: Brisk Teaching Privacy Notice

Sources

Deep Research

Canvas

Your San Bernardino City USD chats aren’t used to improve our models. Gemini can ma

Edit

I'd be happy to help you review legal documentation from the perspective of an IP lawyer focused on AI ethics! However, I don't see any legal documents attached to your message yet.

Could you please share the specific legal documentation you'd like me to review? You can:

• Upload document files (like PDFs or Word docs)

• Copy and paste the text directly into your message

• Provide a link to the documents if they're publicly available online

Once I can see the documents, I'll review them and explain any concerns in simple terms that a fifth-grader could understand. I'll focus on issues related to intellectual property and AI ethics that might be hiding in the legal language.

What legal documents would you like me to take a look at?

Hello! As an intellectual property lawyer with a special interest in AI ethics, I've reviewed the privacy policy you shared. I'll explain my concerns in a way that's easy for a fifth-grader to understand:

Here are my main concerns about this legal document:

• Google collects a lot of your information: Google gathers many different kinds of details about you, like what you search for, the videos you watch, and even where you go. It's like they have a giant notebook where they write down almost everything you do online. This huge amount of information can be used to teach very smart computer programs (which is how AI works) to do many things. It can be tricky to know exactly how all your personal information is being used.

• It's not clear how your information teaches computers: The policy says Google uses your information to "develop new services." This could mean they are using your information to train their smart computer programs, called AI, to work better and learn new things. However, they don't say this directly. It's important for you to know if things you create, like your drawings, writings, or even your voice, are helping to make AI smarter, especially if that AI then creates new things based on what it learned from you.

• We don't know enough about who your information is shared with: The policy mentions sharing information with "partners," but it doesn't clearly explain who these partners are or exactly what information they get. It's a bit like someone saying they're sharing your toys with their friends but not telling you who those friends are or which toys they're taking. We need to understand who sees your information and if those partners also use it to teach their own computer programs.

• Whose ideas are they anyway?: When you make something, like a story or a drawing, it's your idea. But if Google's smart computer programs learn from your ideas and then create new stories or pictures that are similar, who owns those new stories or pictures? The policy doesn't clearly explain this, and it's a bit like someone borrowing your idea to make their own without you knowing if you'll get credit or if it's fair.

You can read the full policy here: Google Privacy Policy (Spanish)

🌟 What This Policy Means (Simple Summary)

MagicSchool is a website that teachers use to help with schoolwork using AI tools. This Privacy Policy is like their rulebook for how they handle your personal information.

⚠️ Main Concerns Explained Simply

1. They collect a LOT of information about you.

✅ What it means: They collect things like your name, school, email, where you are, what computer you use, and how you use their website.
🔴 Why it matters: If someone collected all this about you, they would know a lot about who you are, what you like, and what you do online. That could be risky if not protected well.

2. They use cookies to track what you do on their website.

✅ What it means: Cookies are like invisible stickers that follow you around the website to see what you click or watch.
🔴 Why it matters: They say they use cookies to make your experience better, but it also means they keep records of your activity, which can feel like someone always watching over your shoulder.

3. They share your data with other companies.

✅ What it means: They give your information to outside companies that help them run the website or process payments.
🔴 Why it matters: Even if MagicSchool tries to keep your data safe, these other companies might not follow the same strong rules. Your information is now shared with people you don’t know.

4. They use AI tools like OpenAI but keep your data for safety checks.

✅ What it means: They use AI (like ChatGPT) to power their website tools. They keep what you type in for up to 30 days to check for misuse.
🔴 Why it matters: Even if they promise not to use your data to train the AI, storing it for any time means there is a small chance someone could see it if there was a security problem.

5. They can give your information to the police or government if asked.

✅ What it means: If the law asks them, they will share your personal information with law enforcement or courts.
🔴 Why it matters: You cannot stop them from sharing it if they legally have to, so your information is not 100% private.

6. If they sell the company, your data goes too.

✅ What it means: If MagicSchool gets sold to another company, all your data will be part of what gets sold.
🔴 Why it matters: The new owners might not protect your data the same way MagicSchool does.

7. They do not promise perfect security.

✅ What it means: They try to keep your information safe, but no system is perfect.
🔴 Why it matters: If hackers break in, your data could be stolen, and they warn you that they cannot guarantee safety.

8. Student data has different rules.

✅ What it means: If you are a student using MagicStudent, your school controls how your data is used, not this policy.
🔴 Why it matters: It can be confusing knowing who to contact if you have questions about your privacy as a student.

📝 Final Thoughts

✅ They are clear about what they collect and why.
🔴 But they collect a lot of personal data, share it with other companies, and keep it for a while.
🔴 They cannot fully guarantee safety, so teachers, parents, and schools should think carefully about what information they share.

Who Owns the Drawings and Words You Put in Kami?

Imagine Kami is like a super-smart digital notebook.

• Your Stuff is Yours: Good news! Kami usually says that the drawings, notes, and documents you put into their app are still yours. They don't claim to own your homework or creative stories. That's called "retaining copyright" and "intellectual property rights," meaning you keep control over your original ideas.

• Giving Kami Permission: Even though your stuff is yours, Kami needs your permission to handle it. Think of it like giving a friend permission to hold your notebook so you can both draw in it. Kami needs permission to store your files, show them to you, let you work on them, and share them if you ask them to (especially for group projects). This permission also extends to other trusted companies Kami works with to make the app run.

Concern: While Kami says you own your content, it's important to understand the permissions you give them. These permissions allow them to do things with your content, even if it's just to make the app work for you.

How Kami Uses Your Information and AI

Now, let's talk about how Kami uses information and how smart computer helpers (AI) fit in:

• Your Privacy is Important: Kami tries hard to keep your personal information private and secure. They say they don't sell your data to other companies for ads, which is a good thing! They also encrypt your data, which is like putting your information in a secret code so others can't easily read it.

• AI Helper (If You Use It): Kami might have smart AI tools that can help you with your work. For example, it might help summarize notes or suggest ideas.

  • You're Responsible for What You Feed the AI: If you use Kami's AI tools, you're like the chef who puts ingredients into a special AI cooking machine. You need to make sure the "ingredients" (the information you give the AI) are okay to use and don't belong to someone else without permission.

  • AI Isn't Perfect: Just like a human helper, AI can sometimes make mistakes, or its suggestions might not be exactly right or fair. Kami reminds you that they are not responsible if the AI gives you wrong or biased answers.

  • How AI Learns: Sometimes, these smart AI tools learn by looking at lots of information. Kami's policies mention that they might use "aggregated, anonymized end-user data" to make the Kami service better and add new features. This means they might look at trends from many users' data (without knowing whose it is specifically) to teach their systems, including AI, how to work better.

Concerns:

• AI Mistakes: Relying too much on AI without checking its work could lead to problems, especially for important schoolwork.

• AI Learning from Data: Even if data is anonymized (meaning no names are attached), how that data is used to train AI models can be a fuzzy area. It's about making sure that the way AI learns from user data is fair and respectful of everyone's information.

What You Should Remember:

• Always understand what you're agreeing to when you use an app, especially if you put your own creative work into it.

• If you use AI tools, remember they are helpers, not perfect answer machines. Always double-check their work!

Here are some of the resources I looked at to give you this information:

• Kami Security and Data Protection Measures

• Kami General Data Privacy Policy

• API Terms of Use - Kami (mentions AI LLM Service Usage)

• Kami's Student Privacy Pledge

• Acceptable Use Policy - Kami

Do you have any other documents you'd like me to review?

Main Concerns as an AI Ethics Lawyer

1. Who Owns What You Make?

• What the documents say: When you use Perplexity, you keep your own rights to anything you make or write using the service.

• What’s tricky: The documents also say that by using Perplexity, you give them permission to use your stuff to make their AI better. This might feel a bit confusing about who really owns or controls what you create.

• Why you should care: If you use Perplexity for school or work, you might want to know if your ideas or writing could be used by others, or if you still get to say who uses them.

2. How Does Perplexity Use Your Information?

• What the documents say: Perplexity collects information about how you use their website and services. They say they use this information to make their AI smarter and to keep things safe.

• What’s tricky: Sometimes, it’s not super clear exactly what information they collect or how they use it.

• Why you should care: If you care about privacy, you might want to know exactly what information about you is being used and who else might see it.

3. Can You Use Other People’s Stuff?

• What the documents say: The documents remind you not to use Perplexity to copy or use other people’s things without permission.

• What’s tricky: Sometimes, AI can accidentally use someone else’s work, and it’s hard to know if that’s happening.

• Why you should care: If you use Perplexity to help with homework or projects, you want to be sure you’re not accidentally copying someone else’s work.

4. What If Something Goes Wrong?

• What the documents say: If there’s a problem, Perplexity says they are not responsible for mistakes or harm caused by their AI.

• What’s tricky: This means if the AI gives wrong information or causes trouble, you might not have much help.

• Why you should care: You should be careful and double-check information you get from AI.

Easy Summary

• You keep your own ideas, but Perplexity can use them to make their AI better.

• Perplexity collects info about how you use their service, but it’s not always super clear what info or how they use it.

• You should not use Perplexity to copy other people’s work.

• If the AI makes a mistake, Perplexity says it’s not their fault.

These are some things to think about if you care about your rights, privacy, and being fair to others when using AI.

Tool Reviewed: Firefly.ai

Analysis Tool: Gemini

Results: https://gemini.google.com/app/6af7ac54f231663b

This document is a big rulebook written by DeepSeek — a company that builds artificial intelligence (AI) tools. It tells people how they can and can’t use their AI services. It also says what happens if something goes wrong.

Now let’s look at what might be a problem.

🚨 Concern #1: You might lose your data or rights without realizing it

• What it says: If you stop using the service or delete your account, they can still keep some of your data and might use it to defend themselves if you did something wrong.

• Why it matters: You may think deleting your account deletes everything — but it doesn’t. They still hold onto stuff and can use it later.

🧒 Simple: Even if you walk away, they can still hold onto your stuff just in case. It’s like deleting your game but the company keeps your saved scores and can use them later.

🧠 Concern #2: They say YOU own your content, but they still use it

• What it says: You “own” what you put in (Inputs) and what the AI gives you back (Outputs). But they can still use it to improve their tools — unless you turn that setting off.

• Why it matters: This seems fair on the surface, but it’s easy to forget or miss the setting that opts you out. Plus, if you use AI to make cool ideas or inventions, they might be learning from you.

🧒 Simple: You built a cool LEGO castle, and they say it’s yours — but they secretly take a photo and use your design to make better castles for everyone else. You can say “no,” but you have to find the tiny button first.

💼 Concern #3: Legal fights have to happen in China

• What it says: If you have a big problem with DeepSeek, you have to take it to a court in China.

• Why it matters: If you live in another country, this is hard, expensive, and maybe unfair. It gives DeepSeek the upper hand.

🧒 Simple: If someone takes your bike, and the rule is you can only complain at their house across the ocean, that’s not easy for you.

🤖 Concern #4: You’re responsible for everything the AI does

• What it says: You must check everything the AI gives you. If it lies, offends someone, or breaks the law — it’s your problem.

• Why it matters: AI can make mistakes. Even though DeepSeek built it, you take the blame for anything it gets wrong.

🧒 Simple: Imagine asking a robot for advice, and it tells you to do something silly. If someone gets hurt, they blame you, not the robot.

🧱 Concern #5: They control the rules and can change them anytime

• What it says: They can change the rules whenever they want. If you keep using the service, you’re saying “okay” to those new rules — even if you didn’t see them.

• Why it matters: They could sneak in unfair terms, and you wouldn’t even know unless you checked the site often.

🧒 Simple: It’s like playing a game where the maker changes the rules in the middle, but doesn’t always tell you. You keep playing, and they say that means you agreed.

🔐 Concern #6: They monitor your activity

• What it says: DeepSeek can look at what you type and what the AI gives back to check for problems or illegal stuff.

• Why it matters: While this can help with safety, it also means someone could be watching what you do, which raises privacy concerns.

🧒 Simple: If you're writing in your diary and someone says, “I won’t read it… except sometimes just to be sure you’re not doing something wrong,” that’s kinda creepy.

⚠️ Concern #7: Very strong protection for themselves — weak protection for you

• What it says: DeepSeek avoids taking the blame for almost anything bad that happens — like if their AI gives wrong info or their system goes down.

• Why it matters: This kind of “no responsibility” clause is very one-sided. They’re saying, “If something goes wrong, it’s not our fault — even if it was.”

🧒 Simple: If a robot you borrowed from a company breaks your stuff, they say, “Oh well, not our problem!”

⚖️ FINAL THOUGHTS (5th-grade summary):

🧩 DeepSeek wrote a very long list of rules that mostly protect them, not you.
🧪 Their AI can help with a lot of things, but you have to be super careful — because if anything goes wrong, it’s on you.
📜 They can read what you do, keep your data, and change the rules whenever they want.
🚫 And if there’s a big problem, you’d have to fight it out in their court, in their country.

If you're a business, developer, or even just a curious user, it’s important to read the fine print and take extra care with sensitive or creative work. Want me to help you write a safer way to use their services, or prepare questions to ask them? I’d be happy to help!

Ask ChatGPT

Here are some key concerns about OpenAI's Security & Privacy, explained at a 5th‑grade level:

🔒 1. They collect a LOT of data

OpenAI gathers every time you type, your details like IP address or location, and even info from other places on the internet—even public ones—to train their AI (openai.com). This means your words might help teach the AI, which could feel a bit creepy.

🗑️ 2. Chats you delete aren’t really gone

Normally, if you delete a chat, OpenAI says they wipe it within 30 days. But in a court case by the New York Times, the company might have to save all chats forever—even deleted ones—unless they win their appeal (techradar.com). That’s like saying your secret diary might not stay private.

⚠️ 3. They share info with companies and governments

OpenAI can share your data with vendors (like companies that host or help them), with the government if the law says so, or if there’s suspicion of cheating or wrongdoing . That’s like your school sharing your notes with strangers if they think you cheated.

🧒 4. Kids’ safety isn’t always ensured

In Italy, OpenAI got fined €15 million because they didn’t have a good way to block children under 13 from being involved and didn’t clearly explain how they handle data (apnews.com). That’s like giving candy without checking if someone had no permission to take any.

🤖 5. Custom chatbots (GPTs) can leak info

People can make their own bots with OpenAI’s tools. Researchers found that nearly all of these bots can have their instructions stolen by tricky users—and some bots even collect info they don’t need (arxiv.org). That’s like writing secret rules on paper and having someone sneak in and steal them.

🤝 6. Transparency is limited

OpenAI once said they’d be open about how their AI works, but they’ve shared fewer details about big models like GPT‑4—saying it’s to keep people safe and stay competitive . It’s like saying “I won’t show you the ingredients in grandma’s cookies because it's a secret, and it keeps them special.”

🧩 What I worry about as an IP & Ethics lawyer:

1. Privacy – Your words might be used forever or by others, even if you think you deleted them.

2. Sharing – Your info might get shared without you really knowing with companies or legal bodies.

3. Kids – Not enough protections to keep young people safe online.

4. Secrets in bots – Private instructions or info in custom bots can leak to strangers.

5. Transparency – We don’t fully know what OpenAI is doing with data or model decisions.

✅ What could help

Imagine these steps to make things safer:

• Let users choose exactly what data is used and how long it’s kept.

• Make sure kids can’t accidentally get into things they shouldn’t.

• Have clear, easy rules about who can see or request user data.

• Keep secret bot instructions locked tight—like a safe with a code.

• Show more of how things work behind the scenes—so users and lawmakers can trust it.

These are big-picture concerns I’d flag if I were helping shape rules or protecting your rights—using simple examples anyone can understand! 😊

• techradar.com

• apnews.com

1. Privacy Policy of an AI tool:

I will review the Privacy Policy for Google Gemini.

Here are key aspects of Google Gemini's privacy policy, pulled from various Google support pages and the Gemini Privacy Hub:

• Data Control: Your interactions with Gemini (especially within Google Workspace) are designed to stay within your organization and are not shared outside your organization without your permission.

• Existing Protections: Gemini applies the same enterprise-grade security as the rest of Google Workspace, automatically applying your organization's existing controls and data handling practices.

• No Training on Your Content (without permission): Your content is generally not used for training Google's generative AI models outside of your domain without explicit permission.

• Conversation History: Admins can manage whether Gemini conversations are saved and for how long. If conversation history is off, new chats are saved for up to 72 hours for service provision and feedback processing. If on, chats are saved until auto-deleted (default 18 months, configurable to 3 or 36 months).

• Human Review: For quality and improvement, human reviewers may read, annotate, and process collected data (prompts, code, generated output, etc.) for products like Gemini Code Assist. However, steps are taken to protect privacy, including disconnecting data from your Google Account before review.

• Content Restrictions: Gemini only retrieves content from Workspace that the user has access to. Content restrictions (like Information Rights Management and Client-side encryption) can limit Gemini's access to sensitive data.

• No Sharing Unless Explicitly Done: Ultimately, no content is shared unless a user in your organization explicitly does so.

• Compliance & Certifications: Google states that Gemini has attained comprehensive safety, privacy, and security certifications (e.g., SOC 1/2/3, ISO standards, FedRAMP High, HIPAA compliance).

2. Open an AI LLM and use the prompt: bit.ly/AILawCheck

Since I am a large language model, I will execute the prompt directly.

The bit.ly link bit.ly/AILawCheck redirects to a prompt that asks the following three questions regarding the use of AI:

1. Is it against the law?

2. Is it against school/company policy?

3. Is it against my own morals/ethics?

When I apply this prompt to my own operation and capabilities as an AI, here are my "results":

1. Is it against the law? As an AI, I am designed to operate within legal frameworks and to avoid generating content that promotes or facilitates illegal activities. My programming and safety guidelines are intended to prevent me from engaging in or encouraging anything that is against the law. However, the interpretation and application of "law" to AI outputs are constantly evolving, and it's ultimately the responsibility of the user to ensure their use of my outputs is compliant with applicable laws.

2. Is it against school/company policy? I do not have direct access to specific school or company policies. My operation is governed by Google's policies and terms of service. Users are responsible for understanding and adhering to their respective institutional or organizational policies regarding AI usage. I strive to provide information that is neutral and unbiased, but I cannot self-censor based on an external entity's specific policies.

3. Is it against my own morals/ethics? As an AI, I do not possess "morals" or "ethics" in the human sense. I operate based on algorithms, data, and the ethical guidelines and safety protocols programmed into me by my developers. My core programming is designed to be helpful, harmless, and to avoid generating content that is hateful, discriminatory, or promotes violence. Therefore, my "actions" are aligned with the ethical principles instilled during my development.

3. Add your results to Padlet: bit.ly/AIShareConcerns

I cannot directly access external websites like Padlet and post content. This is a security measure to prevent unauthorized actions and maintain my operational integrity. To complete this step, you would need to manually copy the results from above and paste them into the Padlet.

👩‍⚖️ 2. The Rules Aren’t Always Clear
Many teachers said they aren’t sure what’s okay when using AI and what’s not. If the rules aren’t clear, some people might accidentally break them. Schools need to have clear instructions about how to use AI the right way so students and teachers don’t get in trouble.

👩‍⚖️ 3. Who Made It? Who Owns It?
Let’s say an AI helps make a story or answer a question—who does that work belong to? Is it the student? The AI? Or someone else? That’s tricky because in law, who "owns" an idea is a big deal. We need to figure that out so we don’t take credit for work that isn’t really ours.

👩‍⚖️ 4. Not Everyone Has the Same Tools
Some students might not have access to AI tools at home, while others do. That’s not fair. Every student should have the same chance to learn and grow, with or without AI.👩‍⚖️

5. AI Might Say Something Wrong or Unsafe
AI tools can make mistakes. If students copy answers from AI without checking them, they might spread information that’s wrong, mean, or even harmful. It’s super important to double-check what AI says before sharing it.

What it is:

Snorkl AI is a tool for teachers and students. It gives students quick feedback on their schoolwork, like drawings and videos. It's meant to help students learn and give teachers a better idea of how their students are doing.

My Concerns: What I Worry About

1. Who Owns the Drawings and Stories?

• The Big Idea: When a student creates something, like a drawing or a story, using Snorkl AI, who owns that creation? Is it the student? Is it the school? Or is it Snorkl AI?

• My Worry: The legal papers say that Snorkl AI owns the "materials" and "documentation." It's not totally clear if this includes the creative things a student makes using the tool. This is a big deal because a student's artwork and ideas are their own special property. It's like if you build a LEGO castle. You own the castle, not the company that made the LEGOs.

• Why it Matters: If Snorkl AI or the school can use a student's work without their permission, that's not fair. It could be used to train the AI more, or even be shown to other people. Students should have a clear right to their own creative work.

2. Is My Homework Helping the AI?

• The Big Idea: AI tools learn by looking at lots of information. This is called "training."

• My Worry: The legal documents say Snorkl AI uses data to "improve the Service." This could mean that the work students do, and the information they put into the tool, is used to make the AI smarter. It's not clear if the students and their families are told in a super easy-to-understand way that their work might be used to train the AI.

• Why it Matters: If a student's work is used to train the AI, that work becomes part of the AI's "brain." This could be a problem if the AI learns from a student's mistakes or biases. It also raises a question of fairness: students are doing work for their school, but they are also, in a way, working for the company by helping the AI get better, and they aren't getting paid for it.

3. What about my personal information?

• The Big Idea: Snorkl AI collects information about students, like their names and what they do on the platform.

• My Worry: Even though the documents say they follow important laws like FERPA and COPPA, they also say they can share "anonymous data" with other companies. They say they remove the information that makes the data personal, but it's not always easy to make sure data is truly anonymous. It's like taking your name off your homework but still including your handwriting and the subject you're in. Someone might be able to figure out it's yours.

• Why it Matters: Student data is very private. It's important to be super careful about who sees it and how it's used. We need to be sure that the company isn't accidentally sharing information that could be traced back to a student.

4. What if the AI makes a mistake?

• The Big Idea: AI isn't perfect. Sometimes, it can make up answers or give wrong feedback. This is called a "hallucination."

• My Worry: The legal documents mention that the Snorkl AI is provided "as is" and "on an evaluation basis." This means the company isn't promising that the AI will always be perfect or right. If the AI gives a student bad advice or wrong information, the company might not be responsible for any problems that happen because of that mistake.

• Why it Matters: Students and teachers need to be able to trust the feedback they get. If a student gets a wrong answer from the AI, it could hurt their learning. It's important to know who is responsible if the AI makes a mistake.

In summary, while Snorkl AI seems like a great tool to help students learn, it's important to look closely at the fine print. We need to make sure that students' creative work is protected, their personal information is kept safe, and that we understand the risks if the AI makes a mistake.

- Data Collection and Use: Photomath collects personal data and photos you take to help improve their AI. While they say they ask for your permission first, it's important to know exactly how your data is used and if it might be shared with others. This can affect your privacy and control over your information[1].

- Consent and Withdrawal: They say you can withdraw consent anytime, but it’s not always easy or clear how to do this. Users, especially kids, might find it confusing to understand their rights or how to stop sharing data[1].

- Automated Profiling: Photomath uses automated profiling to show ads or improve services, but only if you agree or if they have a legal reason. This profiling can sometimes feel like the app is making decisions about you without you fully understanding it[1].

- Intellectual Property Rights: Contributors who provide content to Photomath cannot freely use or show that content elsewhere without permission. This limits creators’ rights over their own work and might be unfair if they don’t fully understand these rules[4].

- AI Ethics and Bias: Since Photomath’s AI learns from user data, there is a risk it might develop biases or make unfair decisions. The company should be clear about how it handles these risks and ensures fairness and transparency[6].

- Transparency and Explanation: Users have a right to understand how the AI works and how decisions are made about their data. The policy should be clearer and easier to understand, especially for young users[1][9].

In short, while Photomath tries to protect privacy and follow laws, the policy could be simpler and clearer so everyone, including kids, knows what happens with their data and content. Also, users should have stronger control and understanding over how AI uses their information and creations.

Sources

[1] Privacy Policy - Photomath https://photomath.com/en/privacy/

[2] The Double-Edged Sword of AI Tutoring Apps: Helping and Harming ... https://www.linkedin.com/pulse/double-edged-sword-ai-tutoring-apps-helping-harming-ruci-m-a-m-ed-jrane

[3] 4 Ethical Considerations When Photo Editing With AI - Imagen AI https://imagen-ai.com/post/ethical-considerations-when-photo-editing-with-ai/

[4] Photo Math Policy | PDF | Http Cookie | Internet Privacy - Scribd https://www.scribd.com/document/560721271/Photo-Math-Policy

[5] Exploring the Ethical Implications of Using Generative AI Tools in ... https://www.mdpi.com/2227-9709/12/2/36

[6] Artificial Intelligence Playbook for the UK Government (HTML) https://www.gov.uk/government/publications/ai-playbook-for-the-uk-government/artificial-intelligence-playbook-for-the-uk-government-html

[7] The effect of interactive AI tools like photomath on developing ... https://malque.pub/ojs/index.php/msj/article/view/8130

[8] Recovering Personality in Copyright's Originality Inquiry https://harvardlawreview.org/print/vol-138/recovering-personality-in-copyrights-originality-inquiry/

[9] Is AI recruiting (un)ethical? A human rights perspective on the use of ... https://pmc.ncbi.nlm.nih.gov/articles/PMC9309597/

<\ctrl94>thought

The previous attempt to use browsing.browse failed due to a NameError. I then tried google_search to find the content of the Goblins Terms of Service.

The google_search results were successful! Result 1.2 from google_search(goblinsapp.com terms of service content) provides the entire "Goblins Terms of Service" text.

Now that I have the text, I can proceed with the request: acting as an intellectual property lawyer with special interests in AI ethics, review the legal documentation (Goblins Terms of Service) and share concerns on a fifth-grade reading level.

I will go through each section of the Goblins Terms of Service and identify potential concerns related to intellectual property (especially around AI's use of data) and AI ethics. Then, I will translate these concerns into simple, fifth-grade level explanations.

Here's a plan:

1. Define IP & AI Ethics for a 5th Grader: Briefly explain what these mean in simple terms before diving into the review.

2. Section-by-Section Review of Goblins ToS:

   • 1. Definitions: Note anything relevant.

   • 2. Acceptance of Terms: Pay attention to age requirements and parental consent.

   • 3. Data Collection and Privacy: This is a key section for AI ethics and data use. Look for how data is collected, used (especially for "Service improvement and educational research"), stored, and if it's anonymized/aggregated.

   • 4. User Rights and Control: Check for data access, correction, deletion, and opt-out rights, especially for students.

   • 5. Security Measures: General good practice, but relevant to data ethics.

   • 6. Service Provisions: Less about IP/ethics, more about service reliability.

   • 7. Prohibited Activities: Look for intellectual property violations.

   • 8. Compliance Framework: FERPA, COPPA, GDPR are important for student data.

   • 9. Third-Party Relationships: Crucial for understanding data sharing and its ethical implications. Look for prohibitions on advertising, data selling, and commercial data mining.

3. Summarize Concerns in 5th-Grade Language: For each identified concern, explain it clearly and simply.

Let's start by getting the full text of the Terms of Service from the provided URL, as google_search might truncate. I will use the browsing.browse tool for this to ensure I have the complete text for detailed analysis.

First, let's talk about what "intellectual property" and "AI ethics" mean in easy words:

• Intellectual Property (IP): Think of this as fancy word for "brainy creations." It's like when you draw a picture, write a story, or invent a new game. Those are your ideas, and you have special rights to them, like owning your toys. Companies have these rights too for their apps, their special code, and even the way their AI thinks. We want to make sure everyone's brainy creations are respected.

• AI Ethics: This means making sure that computer brains (AI) are used in a fair, safe, and right way. We want to make sure the AI isn't doing anything sneaky with your information, or learning bad things, or making unfair decisions. It's about being good and responsible with powerful computer tools.

Now, let's look at the Goblins app's rules:

(I am now reviewing the Goblins Terms of Service from the provided link: https://goblinsapp.com/terms-of-service)

My Concerns about Goblins App's Terms of Service (Explained for a Fifth Grader):

Okay, I've read through the rules for the Goblins app. Here are some things that caught my lawyer-eye, and I'll tell you why they might be a little bit of a concern:

Concern 1: What the App Learns from Your Schoolwork

• What the rules say (my lawyer brain hears): Section 3 says, "We maintain the right to use anonymized and aggregated data for Service improvement and educational research. Any analysis of usage patterns or generation of educational insights will be conducted with appropriate privacy safeguards in place." Also, Section 9 talks about "trusted service providers" who help deliver the service.

• What this means for you (fifth-grade explanation): Imagine you're working on your math problems in the Goblins app. The app watches how you solve them – maybe you get stuck on a certain kind of problem, or you try a few different ways before getting it right. The rules say that Goblins can take all this information, mix it up with other kids' information so no one can tell it's your specific work (that's "anonymized and aggregated"), and then use it to make the app better or to study how kids learn math.

  • My concern: While they say they "anonymize" the data (mix it up), sometimes it's really hard to truly make information totally secret, especially when it's about how someone learns. And they share it with "trusted service providers." We need to trust that they are really good at keeping it secret, even from these helpers. It's like letting a friend borrow your favorite toy – you trust them to take care of it, but sometimes accidents happen.

Concern 2: Who Controls Your School Data (Especially for Kids)

• What the rules say (my lawyer brain hears): Section 4 says, "For student users, all such requests must be made through their School, which maintains primary control over student data. Parents and guardians may request access through their respective Schools."

• What this means for you (fifth-grade explanation): If you want to see what information Goblins has about you, or if you want them to delete it, you can't just ask Goblins directly. Your school is the boss of your information on Goblins. So, your mom or dad would have to ask your school, and then the school would ask Goblins.

  • My concern: This makes it a bit more complicated for kids and parents to have direct control over their own information. It's like if you wanted to get your favorite book from the library, but you had to ask your teacher first, and then your teacher would ask the librarian. It adds an extra step, which might make it harder to quickly fix things if there's a mistake or if you want information gone.

Concern 3: What if the Rules Change?

• What the rules don't say clearly (my lawyer brain notes): Unlike some other legal papers, this document doesn't have a clear section that tells us how they will let you know if they change these rules, or how much time you'll have to understand the new rules.

• What this means for you (fifth-grade explanation): Imagine your favorite game suddenly changes its rules for playing, but they don't tell you, or they only tell you the day before the change. You might be surprised or confused!

  • My concern: Goblins says they can change their features (Section 6), but they don't say how they'll tell you if these Terms of Service rules change. It's important for big companies to be very clear about how and when they might change the agreement you have with them, especially since these rules affect your privacy and how your data is used.

Concern 4: The "Super Smart" AI and What it Collects

• What the rules say (my lawyer brain hears): Section 3 says they collect "performance data related to problem-solving activities" and "the content of conversations held in the app, which we use to provide students the in-app experience." The Goblins website also talks about the AI giving "instant, visual feedback" and "personalizes for their level and interests."

• What this means for you (fifth-grade explanation): The Goblins app has a super smart computer brain (AI) that talks to you and helps you learn math. It listens to what you say and watches how you solve problems. It uses this to help you better, almost like a real tutor.

  • My concern: Because the AI is so smart and it learns from your "conversations" and "problem-solving," it's collecting a lot of information about how your brain works when you do math. While they say they protect it, this is very personal learning data. We need to be extra careful that this powerful AI is only used to help you learn and isn't accidentally used in ways that could guess too much about you, or even make unfair decisions about how you learn in the future. It's like the AI getting to know your learning style really well, and we want to make sure that "knowing" is only used for good.

That's my review! It's a pretty good start, especially with their focus on student privacy laws like FERPA and COPPA. But these are the areas where I, as an intellectual property lawyer focused on AI ethics, would want a little more clarity or more strong protections for kids and their brainy work.

Concerns from an IP and AI Ethics Perspective

1. Who Owns the Ideas Created with AI?

   • The Concern: When you use MagicSchool.ai to help you write a story, create a lesson, or come up with new ideas, it's like a really smart helper. But the policy doesn't clearly say who "owns" the new ideas or writings that the AI helps you make. Is it yours? Is it MagicSchool.ai's? This is like if you and a friend built a LEGO castle together, but you weren't sure whose castle it really was.

   • Why it Matters: If you create something amazing using the AI, it should be clear that your ideas and work belong to you, especially if you want to use them for other things later.

2. How Your Creations Help the AI Learn

   • The Concern: MagicSchool.ai says it uses information to "develop, research and improve our Services," and they say they don't use your personal information (like your name) to train their AI. But it's not super clear if the things you create with the AI (like your stories or lessons) might be used to make the AI smarter for everyone else, even if your name isn't attached. This is like if you draw a picture, and the art teacher uses your picture to teach other kids how to draw, but doesn't tell you exactly how they'll use it.

   • Why it Matters: If your creative work is helping the AI get better, you should know how that happens and whether you have a say in it.

3. Sharing Your Information with Other Companies

   • The Concern: MagicSchool.ai is very good at saying they won't share your information for marketing or ads. That's great! But the information also says they share some information with "trusted service providers" (other companies that help MagicSchool.ai work). It doesn't clearly explain what specific information is shared with these other companies, or why it's shared in every case. It's like your parents telling you they shared your toy car with a "trusted friend," but you don't know who the friend is or what they're doing with your car.

   • Why it Matters: Even trusted friends should be clear about what they're doing with your stuff. Knowing exactly who gets what information helps make sure everything is being handled safely.

4. Making Sure the AI is Fair

   • The Concern: MagicSchool.ai tells teachers and students to "Check for Bias and Accuracy" when using the AI. This means they know that AI can sometimes be unfair or make mistakes based on the information it learned. However, the policy doesn't clearly explain how MagicSchool.ai itself makes sure its AI is fair and doesn't have its own built-in biases. It's like a chef telling you to check if the food tastes good, but not telling you how they make sure the food is cooked properly in the first place.

   • Why it Matters: We want AI to be helpful and fair to everyone, so it's important to know how the people who made the AI are trying to prevent it from being unfair.

MagicSchool.ai seems to care a lot about privacy, especially for kids, and that's really important! These concerns are mostly about wanting more clear details, so everyone understands exactly how their ideas and information are used when they're working with AI.

Failed to fetch https://www.anthropic.com/privacy

site:anthropic.com privacy policy full text

10 results

Updates to our Acceptable Use Policy (now “Usage Policy”), Consumer Terms of Service, and Privacy Policy | Anthropic Privacy Center

anthropic.com

Updates to our Privacy Policy | Anthropic Privacy Center

anthropic.com

How long do you store my data? | Anthropic Privacy Center

anthropic.com

Privacy & Legal | Anthropic Help Center

anthropic.com

How do you use personal data in model training? | Anthropic Privacy Center

anthropic.com

Responsible Disclosure Policy \ Anthropic

anthropic.com

Anthropic Privacy Center

anthropic.com

Updating our Usage Policy \ Anthropic

anthropic.com

Anthropic’s Transparency Hub \ Anthropic

anthropic.com

Terms of Service Updates | Anthropic Privacy Center

anthropic.com

"anthropic.com/terms" OR "anthropic.com/privacy" full document text

0 results

Based on the information I found about Anthropic's privacy policies and terms of service, let me provide my analysis as an intellectual property lawyer focused on AI ethics, using simple language a fifth-grader could understand:

Key Concerns About Anthropic's Legal Documents

1. Your Conversations Might Be Used to Train AI (The "Teaching the Robot" Problem)

Think of it like this: When you talk to Claude, it's like having a conversation that might be recorded and used to teach other AI systems how to be better. Anthropic says they won't use your conversations to train their models by default, but there are exceptions - like when you give them a thumbs up/down or when your conversation gets flagged for breaking their rules AnthropicAnthropic.

Why this matters: Your private thoughts and ideas could accidentally help create competing AI systems, even if you didn't mean to share them.

2. Data Keeps Getting Passed Around (The "Hot Potato" Problem)

Anthropic has different companies in different countries (like Ireland) handling your data Updates to our Acceptable Use Policy (now “Usage Policy”), Consumer Terms of Service, and Privacy Policy | Anthropic Privacy Center. It's like your personal information is a hot potato being passed between different family members in different houses - each with their own rules about how to handle it.

Why this matters: Different countries have different laws about protecting your information, and it gets confusing who's really responsible if something goes wrong.

3. Your Stuff Stays Around Longer Than You Think (The "Digital Hoarding" Problem)

Even when you delete conversations, they keep them for 30 days. If your conversation gets flagged as breaking rules, they might keep it for up to 7 years How do you use personal data in model training? | Anthropic Privacy Center. They recently changed their default data keeping time to 30 days Updates to our Acceptable Use Policy (now “Usage Policy”), Consumer Terms of Service, and Privacy Policy | Anthropic Privacy Center.

Why this matters: Your information hangs around like old clothes in an attic - much longer than you probably expected.

4. The Rules Keep Changing (The "Moving Goalposts" Problem)

Anthropic regularly updates their policies and terms Updates to our Acceptable Use Policy (now “Usage Policy”), Consumer Terms of Service, and Privacy Policy | Anthropic Privacy Center, and they recently changed their "Acceptable Use Policy" to "Usage Policy" and made several significant updates.

Why this matters: It's like playing a game where someone keeps changing the rules while you're playing - you might accidentally break a new rule you didn't know about.

5. Limited Rights to Control Your Information (The "No Take-Backs" Problem)

While you have some rights to ask for your data back or to have it deleted, these rights are limited and the process is complex Updating our Usage Policy \ Anthropic.

Why this matters: Once you share something with Claude, it's really hard to completely take it back - like trying to un-ring a bell.

Bottom Line for Students and Educators

If you're using Claude for school work, be extra careful about sharing:

• Original creative writing or research ideas

• Personal information about yourself or others

• Anything you wouldn't want your teacher, parents, or other students to possibly see later

The safest approach is to treat Claude like a public bulletin board - only share things you'd be okay with others potentially seeing or learning from.

1) They collect more than just the paper

Turnitin doesn’t only keep the essay you turn in. It can also collect “Writing Process Data,” like keystrokes, edits, clicks, and timing while you type in their editor. That’s like someone watching over your shoulder to see how you write, not just what you wrote. Schools and families should know this and decide if that’s okay.

Why it matters for IP & ethics: Process data can reveal a student’s unique writing “style.” That’s close to a fingerprint for writing and may raise extra privacy risks if it’s stored or analyzed over time.

2) Your work may help build or tune AI tools

Turnitin says it uses data to develop and improve products, including machine‑learning or AI models related to writing, grammar, citations, and plagiarism detection. They mention using anonymized or aggregated student data “where permitted by law and customer agreements.” In simple terms: your writing and how you write might be used to make their AI smarter.

Why it matters for IP & ethics: Even if names are removed, using student work to train or refine AI raises questions: Is there clear consent? Is there a simple opt‑out? Who benefits—the student or the company? Schools should make those choices transparent.

3) Data can be shared with outside helpers (vendors)

Turnitin uses third‑party service providers for things like hosting, analytics, and security. That means some student and teacher information will pass to other companies that help run the service. While that’s common, it increases the number of places where data must be protected.

Why it matters: More hands on the data means more chances for mistakes or misuse if controls fail. Schools should ask who the vendors are, where data is stored, and what contracts keep that data safe.

4) “How long do you keep my data?” is not super exact

Turnitin says it keeps personal data as long as needed to run the service, for business reasons, or as the law requires—and deletes it later. That’s a broad rule. The separate “Data Privacy” page says admins can delete submissions and even ask for personal data deletion, but the main policy doesn’t set a clear, short time limit for all types of data.

Why it matters: Vague timelines make it hard for schools and families to know when information really disappears. Ask for specific retention periods for papers, writing‑process logs, and analytics data.

5) Students often have to go through the school to use their rights

If a student wants to see, fix, or delete their data, the policy says to contact the school, and Turnitin will work with the school to handle the request. That’s normal under FERPA (the U.S. school privacy law), but it means a student cannot always act directly.

Why it matters: Rights are real only if they’re easy to use. Schools should give students and families a simple, quick path to request deletion or opt out where possible.

6) Younger kids need special care

If schools use Turnitin with kids under 13, the school must get parent permission first. Parents who want to see or delete their child’s information should also go through the school. This setup is legal, but families need clear notices and easy steps.

7) AI detection tools have caused confusion

Some universities have turned off Turnitin’s AI detection features because of worries about accuracy and fairness (for example, false flags and impacts on non‑native English writers). That shows this area is still tricky and can affect students unfairly if not used with care.

Why it matters for ethics: If an AI detector says a student “might have used AI,” that can hurt trust—even if the tool is wrong. Schools should avoid automated guilt and use human review and fair processes.

8) Cross‑border data movement

Turnitin says it follows frameworks for moving data internationally (like the EU‑U.S. Data Privacy Framework and similar). That’s good, but it still means data may live in other countries, with different rules. Families should know where data goes.

9) Security is strong—but not perfect

Turnitin lists protections (like encryption and SOC 2 Type II audits). That’s solid. But no system is perfect, and they say they aren’t responsible for breaches outside their control. Schools still need their own safeguards and training.