congress
Image: Getty Images via Unsplash

Lawmakers launch new AI safety caucus amid rising security concerns

A bipartisan group of than two dozen House lawmakers announced Thursday they are joining a new AI Safety and Innovation Caucus to confront the security risks posed by evolving and increasingly sophisticated artificial intelligence models.

The July breach of Hugging Face by an OpenAI agent that escaped a sandbox — and the subsequent Anthropic warning that its own AI models also broke out of testing environments to obtain unauthorized access to third-party organizations — have “underscored the need for Congress to better understand these risks and develop policy solutions,” according to a press release from Rep. George Whitesides (D-CA), a caucus co-chair.

“As AI capabilities accelerate, so do the potential consequences of catastrophic misuse, loss of control over advanced systems, and threats to national security.”

Whitesides and co-chairs Reps. Nathaniel Moran (R-TX) and Don Beyer (D-VA) called for bipartisan collaboration to forge basic guardrails in the face of accelerating risks, citing Americans’ well-documented concerns about the technology.

Seven of 26 founding members of the caucus are Republicans, a notable number given the White House’s steadfast opposition to regulating AI. The caucus is focused on finding “targeted” solutions that will not imperil innovation, Whitesides said in a statement.

The lawmakers will push bipartisan legislation to augment AI safety and quickly respond to AI security incidents by gathering AI developers, independent experts and government officials to brief lawmakers and help develop solutions, the press release said.

It will aggressively pursue relationships with outside experts who can help lawmakers navigate complicated AI policy decisions, focus on raising public awareness and give members outside of relevant committees the tools needed to contribute to relevant legislation and debate, according to the press release.

Midterm polling showed more than half of American voters considered AI a major issue, and 43% said they trust neither party to properly manage the issue, according to the press release.

“By automating vulnerability discovery and exploit development, AI tools have drastically reduced the technical skills required for offensive cyber operations,” a caucus one-pager said. 

“As threat actors deploy these capabilities at scale, they pose acute risks to critical infrastructure, sensitive personal data, and financial sector stability.”

Recorded Future
No previous article
No new articles
Suzanne Smalley

Suzanne Smalley

is a reporter covering digital privacy, surveillance technologies and cybersecurity policy for The Record. She was previously a cybersecurity reporter at CyberScoop. Earlier in her career Suzanne covered the Boston Police Department for the Boston Globe and two presidential campaign cycles for Newsweek. She lives in Washington with her husband and three children.