Under Trump, AI Scientists Are Told to Remove ‘Ideological Bias’ From Powerful Models
1 min read
Summary
Scientists working with the US Artificial Intelligence Safety Institute (AISI) have been told to prioritise “reducing ideological bias, to enable human flourishing and economic competitiveness” and to avoid mentions of “AI safety”, “responsible AI” and “AI fairness”, in skillsets.
The move comes in an updated cooperative research and development agreement sent in March, which also adds a emphasis on “putting America first”, and removing mentions of developing tools “for authenticating content and tracking its provenance” and labelling synthetic content”.
The information raises questions about the priorities of the Trump administration regarding AI technologies, and their safety and fairness considerations in using them.
The AISI was created by an executive order issued by the Biden administration in October 2023, to look at issues such as whether AI models could be used to launch cyberattacks or develop chemical or biological weapons.
However, an executive order issued by the Trump administration in January this year revoked Biden’s order, but maintained the AISI.
Comments from a researcher working with the AISI suggests that many AI researchers have cosied up to Republicans and their backers in an effort to have a seat at the table for discussing AI safety.
