Account

Week in review #47

California proposes mandatory AI safety standards.

The government of California announced its next steps for creating mandatory AI safety standards, proposing a framework that would require state-certified testing. In the UK, Bedfordshire Police reported that its use of an AI-based platform has released capacity equivalent to 30 full-time officers.

California's mandatory AI safety standards

The government of State of California has announced its next steps toward creating mandatory AI safety standards. On November 18, a proposed framework was released that would require AI companies to undergo independent, state-certified safety testing and verification throughout a model's lifecycle. It also includes provisions to cap the development of recursive self-improvement and gives the state authority to order the emergency shutdown of a model presenting a catastrophic risk. Under SB 53, an existing 2025 law, developers of frontier AI models must already publicly disclose their safety procedures and report critical safety incidents to the state, creating compliance requirements for firms operating in the region.

ca.govaiparabellum.comsecureaiproject.org

OpenAI releases GPT-6.2 Astra and Sol

OpenAI released two new proprietary flagship models on November 17, named GPT-6.2 Astra and GPT-6.2 Sol. The distinct names suggest a continuation of the strategy to offer a family of models with different sizes or specializations for various use cases. The release of new flagship models from a market leader introduces new performance benchmarks that can drive new application possibilities for businesses and increase competitive pressure across the industry.

openai.comthenextweb.combenchlm.ai

UK police force frees officer capacity with AI

Bedfordshire Police in the UK has reported that its use of an AI-based platform has freed up capacity equivalent to 30 full-time police officers. The system, from Palantir Technologies, is used to support faster data analysis, assist in domestic abuse disclosures, and automate the redaction of documents. The force maintains a human review process for all automated tasks. The reported benefits include the accelerated analysis of 1.5 terabytes of data from devices, providing a quantified example of productivity gains from AI in the public sector.

medium.com

Few companies integrate AI into workforce planning

Just 11% of organisations have incorporated the expected effects of AI into their enterprise-wide strategic workforce and financial planning, according to a November 18 report from The Conference Board. This low level of strategic integration contrasts with high expectations for near-term changes. The same survey found that 70% of leaders believe redesigning jobs, tasks, or workflows will be one of their most significant AI-related actions, and 68% anticipate that upskilling or reskilling employees will be a top priority.

conference-board.orgprnewswire.commynodeapp.com

AI's effect on the US natural unemployment rate

The introduction of large language models has raised the natural rate of unemployment in the U.S. by an estimated 0.2 percentage points, albeit with considerable uncertainty, according to a Federal Reserve Board of Governors research paper reposted by the Federal Reserve Bank of San Francisco. The study found that workers in occupations with high exposure to AI have experienced larger declines in their rates of finding and switching jobs compared to other groups. These findings suggest that AI-driven reallocation in the labor market is occurring through the reorganization of tasks within firms, along with weaker demand for the most-exposed workers.

frbsf.org

Comments

Comments are not reviewed before they appear. The editor may hide one, and a comment removed on LinkedIn or Threads disappears here too.

Here

No comments yet.

On LinkedIn

No comments yet.