Anthropic CEO urges AI slowdown; Fresno State, UC Merced weigh safety steps
Dario Amodei says the AI industry should slow to let safety catch up. Central Valley campuses are building oversight as California tightens rules.
Anthropic CEO urges AI slowdown; Fresno State, UC Merced weigh safety steps
Key Takeaways
- Anthropic CEO Dario Amodei urged slowing AI development to allow safety measures to catch up.
- He proposed embedded third‑party evaluators with employee‑like access at AI firms.
- California just enacted laws to create AI model risk evaluations and an auditor registry.
- Fresno State and UC Merced have formal bodies focused on AI policy and responsible use.
"If slowing down bought us even an extra year or two before models reach critical levels of capability, we could greatly reduce the risk," Anthropic CEO Dario Amodei said. The call lands in the Central Valley as UC Merced and Fresno State formalize AI oversight, and as California moves to new safety rules that could shape campus use and local employers’ adoption. UC Merced’s AI Advisory Council is chaired by Avi Badwal.
It was 97 degrees in Fresno on Sunday afternoon. For now, caution.
What Amodei is asking for
Amodei urged the industry to slow development so safety can catch up, warning that within six to twelve months AI could coordinate harmful actions online if left unchecked. He outlined steps including giving independent evaluators ongoing, employee‑like access, with desks and badges, to monitor safety practices inside AI companies. OpenAI’s Sam Altman publicly backed part of that plan. "We owe it to humanity to try," Amodei wrote.
Amodei’s warning followed a turbulent week for the sector, with public resignations by AI researchers and fresh disclosures about attempts to misuse advanced models for cyberattacks and surveillance. Anthropic has said it disrupted some of those efforts.
What California just did
On Thursday, Sept. 10, Gov. Gavin Newsom signed a package targeting tech risks to kids, and he also approved two measures aimed at AI oversight that Anthropic supported. One directs the state to set rules for independent organizations to evaluate AI model risks. The other creates a registry of financially independent AI auditors. The timing puts California closer to the kind of third‑party review Amodei described.
What this means for Valley campuses
Fresno State has stood up an AI Initiative under the Office of the President, including an AI Policy, Ethics, and Governance Committee, and names executive sponsors such as Vice President and CIO Dr. Bao Johri. The university says its effort focuses on ethical guidance, workforce development, and campus‑wide policy. That structure gives Fresno State a way to absorb any state‑level rules about model evaluations and auditors.
UC Merced’s AI site lays out responsible‑use principles tied to University of California guidance and warns against sharing sensitive data with unsupported tools. Its AI Advisory Council, chaired by Badwal, lists workgroups on teaching, research, and administrative services, and includes faculty such as Stefano Carpin and Josh Viers among the leads. The council’s charge includes advising on ethics, privacy, and responsible usage.
The Central Valley question
For students, city staff, and ag firms testing automation tools, the practical question is simple. If California moves to more formal third‑party evaluations, which tools clear that bar, and who pays for audits. Valley institutions are already writing their playbooks, which means local adoption could continue but with more paperwork and outside eyes.
"The measures I propose to advance the frontier at a safe pace will not be easy," Amodei said. "But I believe we owe it to humanity to try."
Central Valley AI is produced by the CVAI Newsdesk team and developed by Kaweah Tech, a regional firm that builds, deploys, and integrates AI solutions for businesses across California's Central Valley.