Donald Trump’s government has begun to reconsider its more flexible stance on artificial intelligence (AI) following the advancement of systems capable of identifying hidden flaws in computer codes. The movement gained strength after the presentation of the Mythosan Anthropic model aimed at detecting security vulnerabilities.
According to the The Washington Post members of the White House discuss possible measures to deal with the risks associated with the new generation of AI tools. Among the options analyzed is a possible executive order to expand supervision over the sector.
Government discusses new forms of control
The director of the National Economic Council, Kevin Hassett, compared the government’s possible approach to the testing model of the FDA, the United States’ drug regulatory agency. The idea would be to ensure that systems are considered safe before reaching the public.
After the statement, White House Chief of Staff Susie Wiles stated on X that the government does not intend to “pick winners and losers”, indicating that there is still no internal consensus on the issue.
According to an authority spoken to by the publication, the government’s objective is to buy time to deal with risks that could grow as AI models become more powerful.
Mythos raised concerns about security
Anthropic announced last month that Mythos would be able to locate security flaws in software in a highly efficient way. The company decided not to release the system to the general public as it considered the risks to be high.
Shortly after, OpenAI stated that its latest models have similar capabilities. The allegations were supported by assessments from the AI Security Institute, linked to the British government.
According to the report, the Trump administration brought together executives from large banks and representatives from the main AI companies to discuss the impacts of the technology. One of the meetings involved Susie Wiles and Anthropic CEO Dario Amodei at the White House.
Government has already expanded initiatives linked to AI
While evaluating new measures, the government has also begun to strengthen oversight mechanisms. This week, the Center for AI Standards and Innovation expanded a program created during the Joe Biden administration to test models before public launch.
Additionally, the General Services Administration presented a contractual proposal that expands control over AI systems used in federal activities, including the analysis of responses considered ideological.
Source: www.olhardigital.com.br
Source link
