Former Researcher Warns AI Safety Checks Are Too Slow
A former Anthropic researcher argues that the pressure to ship products has left little room for critical safety analysis, creating a dangerous gap in oversight.
Jacob Coxon, who recently left Anthropic, has raised urgent concerns about the pace of artificial intelligence development. Speaking on ABC’s This Week, Coxon stated that employees at major AI firms often lack sufficient time to properly evaluate the risks associated with the models they are building. He described the work environment as one where developers are focused intensely on coding and technical execution, leaving little bandwidth for broader safety considerations.
Coxon, who also previously worked at OpenAI, characterized the current trajectory of the industry as a high-stakes gamble. He suggested that while some employees understand the potential civilizational dangers, the competitive pressure to release capabilities quickly often overrides caution. This tension between rapid innovation and responsible deployment is now drawing significant attention from both industry leaders and political figures.
Safety concerns overshadow rapid development
According to Coxon, the primary focus for engineers is often on solving immediate technical problems rather than analyzing long-term societal impacts. He noted that at companies like Anthropic, the stakes are well understood, but the belief that competitors will not act responsibly drives a race to the finish line. This dynamic means that safety protocols are frequently treated as secondary to the goal of being first to market.
The former researcher emphasized that this approach carries significant risks. He argued that the industry is moving too fast for robust oversight mechanisms to keep up, creating a scenario where dangerous capabilities could be deployed before adequate safeguards are in place. This perspective challenges the common industry narrative that progress is inherently beneficial without sufficient pause for reflection.
Lawmakers urge immediate regulatory action
Political figures have responded to these warnings by calling for decisive action. House Minority Leader Hakeem Jeffries criticized the current legislative approach, stating that the delay in addressing AI risks is a failure of responsibility. He urged his colleagues to pass safeguards immediately, rather than waiting for a future crisis to force their hand.
The political discourse highlights a growing frustration with the current pace of technological change. Lawmakers argue that without clear regulations, companies may continue to prioritize speed over safety. This calls for a new framework that mandates thorough risk assessments before new AI models are released to the public.
Industry leaders consider slowing the pace
Interestingly, Anthropic CEO Dario Amodei has echoed some of these concerns. In a recent essay, he argued that the industry must slow down the rate of capability advancement to allow risk prevention measures to catch up. Amodei suggested that while progress will still appear fast, taking deliberate pauses is necessary to ensure safety.
Coxon viewed Amodei’s stance as a substantial step forward, indicating a potential shift in how top executives view their obligations. He also suggested that American companies should collaborate with their Chinese counterparts on risk prevention, pointing to the global nature of the challenge. This internal and external call for caution suggests a possible recalibration of the industry’s aggressive expansion strategy.
