Anthropic silently, Biden-ERA AI policy obligations from the website

Safe and “Reliable” on the website of the Anthropic site Safe and “Valid” had several voluntary liabilities in 2023 with the Administration of Biden.

Liabilities sharing information on AI risks and AI bias and discrimination and discrimination information management information management information management have been deleted from anthropic Transparency Hub last week, according to To the AI ​​Watchdog group of the Midas project. Other Biden-period obligations Related to the reduction of the form-based sexual exploitation of AI created to stay.

It seems that there is no warning about anthropic change. The company immediately did not meet a commentary request.

Together with companies including anthropic, Openai, Google, Microsoft, Meta and Intrection Declared in July 2023 Biden agreed to comply with certain voluntary AI security obligations offered by the management. Protecting sensitive EU information on liabilities and the AI ​​generated content is included in cibers before investing in cyberscase, including liabilities, and the internal and external security tests of AI systems.

To be clear, anthropic had already accepted a number of experiments shown in liabilities and Accord was not legally binding. However, the intention of Biden’s leadership was to signalize the AI ​​policy priorities in a more perfect way AI Executive OrderA few months later entering into effect.

The Trump Office said that the approach of the AI ​​management will be completely different.

In January, President Trump, National Standards and Technology in the Author and Models, which helped the Institute of National Standards and Technology, impartial. Trump with Trump, critics, defended the statements of the order of the order of the order, in a difficult and effective trade secrets of commercial companies.

Shortly after the EU canceled the execution order, Trump, “Dreaming of human prosperity, economic competitiveness and economic security” signed the order of Directing Directing to promote the development of the EU. The important thing, the order of the Trump did not create any words about the fight against AI discrimination, which is the main teneti of Biden’s initiative.

As a Midas Project note In a number of posts in a number of posts in X, liabilities in a number of posts, the time of the word or the president of the seating did not claim that time was closed or contingent. In November, more than one EU companies after the election confirmed Their obligations have not changed.

Anthropic is not the only company to adjust its open policies in months since the task of Trump. Openai recently announced cuddle “Intelligent freedom … how difficult or controversial topic can be” and Try to ensure that AI does not provide certain points censor.

Openai too He rubbed a page on the website The start is used to express the diversity, capital and input or dei obligation. These programs are fired from Trump management, who led a number of companies to eliminate or eliminate SEI initiatives.

The EU claims to include Trump Silicon Valley Consultants, including Marc Andreessen, David Sacks and Elon Musk, including Google and Openai AI is engaged in restricting the answers to AI Chatbots with censorship. Laboratories, including Openai, rejected policy changes in response to political pressure.

Actively implements both Openai and anthropical and anti-anthropical and public agreements.

A few hours after this story was published, anthropic TechCrunch sent the following statement:

“We are committed to the volunteer AI liabilities created within the Biden administration. These progress and concrete actions continue to be reflected in the center of the Transparency Center. To prevent additional confusion, we will add a section to the place where our progress is adapted.”

UPDATED 11:25 AM Pacific: Added a comment from anthropic.