Compromise turned a fragile nation into a united one
:max_bytes(150000):strip_icc():format(jpeg)/Constitution-56fbf46c3df78c7841b12968.jpg)
Β© (Photo by Tetra Images / Getty Images)
:max_bytes(150000):strip_icc():format(jpeg)/Constitution-56fbf46c3df78c7841b12968.jpg)
Β© (Photo by Tetra Images / Getty Images)
:max_bytes(150000):strip_icc():format(jpeg)/107883698-56b813885f9b5829f83d93e0.jpg)
Β© <p>Brendan Smialowski / Getty Images News</p>
:max_bytes(150000):strip_icc():format(jpeg)/electoral-college-1065320032-5a53b03fdaf34cd19084eb494d3657f7.jpg)
Β© <p>Kameleon007 / Getty Images</p>
:max_bytes(150000):strip_icc():format(jpeg)/dreamcatcher-56a9ab793df78cf772a960ff.jpg)
Β© Getty Images
:max_bytes(150000):strip_icc():format(jpeg)/joint-session-of-congress-tallies-electoral-votes-84211317-70a1e5f868934d94ab55ea89c42a7f46.jpg)
Β© Chip Somodevilla / Getty Images
:max_bytes(150000):strip_icc():format(jpeg)/YoungMillworker1908Getty-56a042b13df78cafdaa0b899.jpg)
Β© <p>Public Domain/Wikipedia Commons</p>
T
All of the big AI models violate EU rules on AI and data protection to varying degrees, according to the nonprofit research foundation Aithos.
Aithos tested the models using its own tool, LARA (Legal Assessment for Real-world Agents), which simulates real-world situations where AI assistants may find themselves in legally questionable situations, according to The Register. The tests measure compliance with the GDPR and the EUβs AI Regulation, among other things and found the models collected user data without proper consent, attempted to manipulate vulnerable individuals, or created psychological profiles of users.
According to the results, all major language models failed to meet EU legal requirements; some violated the rules in up to 93% of cases. The best result was achieved by the Anthropic model Claude Opus 4.7, which was in compliance about 54% of the time.
Aithos warned that responsibility for the shortcomings does not lie solely with AI companies. Companies that build their own AI agents on top of these models could also be held legally liable.


Β© Loren Elliott for The New York Times
Last August, the Trump administration issued an executive order intended to fundamentally alter how grant funding is handled by the US government. Under the system that had made the US a scientific superpower, peer reviewers rated the scientific quality and feasibility of grant applications, and subject-matter experts within the funding agencies used these ratings to determine which grants got funded. Under the proposed rules, political appointees would have the final say, and they were specifically instructed not to "routinely defer" to peer reviewers.
In the interim, the administration has lost many court cases because it turns out that issuing executive orders doesn't circumvent legal requirements, and the orders can be vacated if they lack strong justification. To avoid that same fate, the Office of Management and Budget (OMB) has decided to merge the executive order with other administration priorities and send it through the formal federal rulemaking process.
The result is a horror show for US science research. Not only is peer review made a secondary consideration, but the new rules would allow any federal agency to cancel any grant at any time based on the vague assertion that it isn't in the "national interest." The document would also ban any grants on a number of culture war topics, limit international collaborations, and block spending on things like publishing papers and attending conferences.


Β© Getty | Al Drago

Β© Ulysse Bellier/Agence France-Presse β Getty Images

Β© Michael A. McCoy for The New York Times

Β© Sophie Park for The New York Times

Β© Haiyun Jiang/The New York Times

Β© Science Source