The company says it will issue the report annually to assess progress toward its public-benefit charter.
The report evaluates five pillars: solving technical AI-safety problems; making safety practice universal and legally backed; strengthening democracy and global stability; expanding prosperity and human flourishing; and sharing AI benefits broadly. Anthropic is a Public Benefit Corporation, and the document includes closing commentary from its Long-Term Benefit Trust.
Among the examples Anthropic reports are Claude’s work on protein binders, support for WHO Africa Ebola situation reports, and assistance to cyberdefenders finding more than 33,000 critical- and high-severity flaws. It also says, subject to stockholder approval, it will set aside at least 5% of its shares for projects pursuing its public-benefit charter.
The report is Anthropic’s own assessment. It does not independently verify the cited outcomes, establish that Claude caused them, or provide an external audit of progress against the five pillars.