OpenAI and Anthropic made a special report to the staff of the U.S. House of Representatives Homeland Security Committee on their new generation AI models with network attack and defense capabilities, as well as the various impacts of such technologies on network security. This is one of the first meetings for the U.S. legislative body to conduct special communication with the two major AI giants on the cyber threats caused by new AI models, focusing on the security risks faced by critical infrastructure areas with weak protection.
A committee aide confirmed that the two companies conducted special briefings to congressional staff in two separate, closed and confidential meetings last Thursday.
The assistant said that this communication is a proactive docking between Congress on the latest progress of cutting-edge AI models, focusing on studying and judging its potential impact on the network security of critical infrastructure. The briefing also mentioned a recent memo from the White House, which accused China of launching a special operation on an industrial scale to dismantle and replicate advanced American AI models.
OpenAI responded that in addition to the special meeting at the White House last week, the company had completed multiple similar reports simultaneously with multiple committees of the House and Senate.
An Anthropic spokesperson said that the company has long and regularly explained the capabilities of AI models and potential national security impacts to congressional staff, and last week's meeting was part of normalized and ongoing communication.