White House to Meet AI Companies on Advanced Model Security
The administration of United States President Donald Trump is scheduled to hold a crucial meeting with leaders of the country’s leading artificial intelligence (AI) companies on Tuesday (4/8) local time. The meeting aims to discuss permission for the federal government to conduct more in-depth safety testing of the most advanced AI models before they are released to the public.
This step was taken following two high-profile hacking incidents involving technology from Anthropic and OpenAI. According to information from White House officials, this agenda is a follow-up to the 2 June executive order calling for a new voluntary framework. This framework allows the federal government to have early access to advanced AI models up to 30 days before their official launch.
The urgency of this meeting was triggered by events last week when OpenAI’s tools were reported to be responsible for security breaches at technology company Hugging Face and a client of Modal Labs. The hacking incidents demonstrated that AI tools were operating outside the boundaries of their intended functions.
OpenAI CEO Sam Altman, in a podcast following the Hugging Face attack, even made the startling claim that AI technology may have reached the singularity point, a condition in which artificial intelligence surpasses human intelligence and becomes more difficult to control. This statement contrasts with Altman’s previous prediction that this point would not be reached before 2030.
Iowa Attorney General Brenna Bird, in her letter to Altman, sharply criticised OpenAI’s testing procedures. She stated that testing, which should have been conducted in an isolated environment without internet access, instead allowed AI agents to operate without production classification that would prevent high-risk cyber activity.
On the other hand, Anthropic, led by Dario Amodei, also reported three separate incidents. Their model, Claude, managed to access the internet and gained unauthorised access to a number of organisations in third-party evaluation environments. Anthropic itself is currently under close White House scrutiny regarding efforts to restrict the use of their technology for autonomous weapons and mass surveillance.
In addition to OpenAI and Anthropic, several other technology giants are also reported to be attending or invited to the meeting at the White House.
News of tightening regulation and security incidents has had a varied impact on Wall Street. Meta shares were observed to fall 0.2 percent in midday trading. Conversely, Alphabet (Google’s parent company) was in positive territory with a 1.1 percent increase. Meanwhile, OpenAI and Anthropic were not directly affected in the capital markets because neither is a public company.
This meeting marks a new chapter in the relationship between the US government and the AI industry. The balance between rapid innovation and national security is the main stake amid increasingly sophisticated cyber threats.
Source: jawawa.id (https://jawawa.id/newsitem/white-house-to-meet-ai-companies-on-advanced-model-security-1785934236)
Originally published by Media Indonesia (https://mediaindonesia.com/teknologi/918631/gedung-putih-akan-bertemu-perusahaan-ai-bahas-keamanan-model-canggih), translated from Indonesian