355
Audio & Video Production344
Automation & Workflow224
Software Development250
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics140
Design & Creative169
Customer Support131
Photography & Imaging156
Sales & Outreach125
Voice & Speech135
Operations & Admin87
Education & Learning131
Trump signed an executive order asking AI companies to volunteer for safety tests. Critics say it may not stop risky AI systems from launching.
In short: President Donald Trump signed an executive order that expands voluntary government safety testing for advanced AI models, but critics say it may do little to prevent risky releases.
Trump signed an executive order on Tuesday that encourages AI companies to work with the US government on safety reviews for “frontier” AI models, meaning the most advanced systems.
The order does not require companies to submit their models. Instead, it creates a voluntary process. Think of it like asking car makers to bring new cars in for an optional inspection before they go on sale.
One key change from an earlier draft is timing. A previous version would have tried to give the government up to 90 days of early access to test models. The signed version shortens that window to 30 days, according to Politico.
The order tells the National Security Agency to set up a classified benchmarking process, which is a private scorecard to decide which models count as “covered frontier models.” It also calls for a “cybersecurity clearinghouse,” a central place to find and fix security weaknesses. The NSA is told to work with the Treasury Department and the Cybersecurity and Infrastructure Security Agency (CISA).
Critics say the government may not be ready to do meaningful testing that fast. The order also points to possible funding and hiring gaps, including a 60 day timeline to expand cybersecurity hiring pathways.
Advanced AI can help write code, and that can also help people find ways into computer systems. If testing is optional and rushed, critics warn that powerful models could be released before anyone outside the company has a clear view of the risks.
Source: Arstechnica