OpenAI Asks Congress Whether Coordinated AI Development Slowdowns Would Break Antitrust Law

In brief

• OpenAI has asked lawmakers whether rival AI developers could legally coordinate a slowdown, according to WIRED.
• Experts say funding pressures and intense competition discourage companies from slowing development on their own.
• A bill in Congress would protect certain agreements to restrict AI development for security reasons, subject to advance notice to the Justice Department.

OpenAI’s chief scientist, Jakub Pachocki, has called for voluntary slowdowns while the industry works out shared safety standards. The company recently reached out to congressional staffers to ask whether an agreement among AI firms to pause development could violate antitrust laws, according to people familiar with the matter.

The outreach follows a broader industry debate about the risks of rapid AI advancement. Companies face commercial and geopolitical pressures to release powerful models quickly, even when internal researchers warn that more testing is needed. “Commercial and geopolitical competition in the AI space is incredibly intense, leading to a concerning dynamic where companies are incentivized to release products before their risks are fully understood,” said Miranda Bogen of the Center for Democracy and Technology.

Financial incentives further complicate coordinated pauses. “Every advance under current conditions yields many millions or billions more in funding and puts the creators of that advance in a greater position of power and influence,” noted Duncan Sabien of the Machine Intelligence Research Institute.

Legislative efforts are underway to address these concerns. Senators Adam Schiff and Jim Banks have introduced a bill that would protect certain security collaborations from antitrust scrutiny, provided they are notified to the Justice Department in advance.

The debate comes as the United States and China vie for AI leadership. Recent policy shifts, including a delayed executive order and a voluntary review process for advanced models, underscore the high stakes of the technological race.

Recent high‑profile warnings have amplified the debate. Former Anthropic engineer Jacob Coxon resigned, cautioning that AI development could pose existential risks to humanity. “The people building AI earnestly believe that it could kill us all by the end of the decade,” he wrote on X, noting that many executives privately express fear despite reassuring public statements.

Source link