Claude: Difference between revisions
anthropic dmca claims for sharing information they made public |
→Privacy: Claude/Anthropic also asks for a phone number to link your account to your physical identity/phone number, this is mandatory and can't be skipped sadly |
||
| Line 21: | Line 21: | ||
Anthropic provides two ways of accessing their LLMs: subscriptions and direct API usage. API pricing is simple: you pay for what you use. By paying for a subscription you can get up to 13.5x worth of API usage - a "Max" subscription costs $200 per month but allows users to use up to $3,000 worth of actual API costs [[https://she-llac.com/claude-limits source]]. However paying for a subscription means user is locked into Anthropic's tools only - either their web app or desktop app (claude code). Several talented engineers managed to hijack the behavior of Anthropic's tools and hence use limits of their subscriptions in third-party tools. Anthropic responded by changing their policies and banning any suspicions of account subscriptions being used outside of their first-party applications [[https://x.com/robzolkos/status/2024125323755884919 source]]. Several users on twitter/X complained about being banned even without ever taking part in such activities, namely those, who have used multiple accounts with named subscription on the same computer. For a more detailed (and opinionated) view on Anthropic's lack of transparency and unusual business practices, follow [https://www.youtube.com/watch?v=M-pkXr-qqII this video]. | Anthropic provides two ways of accessing their LLMs: subscriptions and direct API usage. API pricing is simple: you pay for what you use. By paying for a subscription you can get up to 13.5x worth of API usage - a "Max" subscription costs $200 per month but allows users to use up to $3,000 worth of actual API costs [[https://she-llac.com/claude-limits source]]. However paying for a subscription means user is locked into Anthropic's tools only - either their web app or desktop app (claude code). Several talented engineers managed to hijack the behavior of Anthropic's tools and hence use limits of their subscriptions in third-party tools. Anthropic responded by changing their policies and banning any suspicions of account subscriptions being used outside of their first-party applications [[https://x.com/robzolkos/status/2024125323755884919 source]]. Several users on twitter/X complained about being banned even without ever taking part in such activities, namely those, who have used multiple accounts with named subscription on the same computer. For a more detailed (and opinionated) view on Anthropic's lack of transparency and unusual business practices, follow [https://www.youtube.com/watch?v=M-pkXr-qqII this video]. | ||
==== DMCA ==== | ====DMCA==== | ||
Anthropic's most popular product Claude Code is closed-source, meaning the actual code used to make the application is not public. Additionally the distributed version of said application is obfuscated. Obfuscation is a common process used in order to make reading code more difficult, essentially impossible without the use of reverse-engineering tools. In 2025 Anthropic accidentally published "source maps" of the application, aiding in mapping the obfuscated code to its original form. Some developers posted said information online, to which Anthropic responded with DMCA claims[[https://techcrunch.com/2025/04/25/anthropic-sent-a-takedown-notice-to-a-dev-trying-to-reverse-engineer-its-coding-tool/ source]]. Github, a platform for sharing code, keeps a track of DMCA claims and makes them public [[https://github.com/search?q=repo%3Agithub%2Fdmca%20anthropic&type=code relevant claims]]. | Anthropic's most popular product Claude Code is closed-source, meaning the actual code used to make the application is not public. Additionally the distributed version of said application is obfuscated. Obfuscation is a common process used in order to make reading code more difficult, essentially impossible without the use of reverse-engineering tools. In 2025 Anthropic accidentally published "source maps" of the application, aiding in mapping the obfuscated code to its original form. Some developers posted said information online, to which Anthropic responded with DMCA claims[[https://techcrunch.com/2025/04/25/anthropic-sent-a-takedown-notice-to-a-dev-trying-to-reverse-engineer-its-coding-tool/ source]]. Github, a platform for sharing code, keeps a track of DMCA claims and makes them public [[https://github.com/search?q=repo%3Agithub%2Fdmca%20anthropic&type=code relevant claims]]. | ||
====Privacy==== | ====Privacy==== | ||
In order to use the LLM, a person must sign in with an e-mail address or to log in with a Google account.{{Citation needed}} | In order to use the LLM, a person must sign in with an e-mail address or to log in with a Google account, as well as obligatorily verify a smartphone number.{{Citation needed}} | ||
Anthropic may use input and output data from their services to train their AI models. Users are able to opt-out if they want to. However, Anthropic may still collect inputs and outputs that belong to conversations that have been flagged for safety review or content that has been reported by the user. | Anthropic may use input and output data from their services to train their AI models. Users are able to opt-out if they want to. However, Anthropic may still collect inputs and outputs that belong to conversations that have been flagged for safety review or content that has been reported by the user. | ||