AI Regulation & Policy

Today's Stories

1. 1

   AI Agent Governance vs. Observability: What's the Difference?

   AI Agent Governance vs. Observability: What's the Difference?

2. 2

   AI investment priorities are shifting from efficiency to innovation, but business leaders are losing confidence in their ability to scale AI across their organizations.

   Akkodis' report shows that innovation has overtaken efficiency as the primary driver of AI investment, signaling a focus on growth and new business models. However, CTO confidence in scaling AI has dropped significantly, with integration challenges across enterprise systems cited as a major obstacle. For business leaders, this signals a disconnect between investment ambitions and operational readiness. The report identifies agentic AI (systems that make decisions and complete tasks autonomously) as a crucial trend, but its adoption demands new governance frameworks—meaning companies cannot simply deploy these systems without rethinking oversight and controls.

   The gap between investment enthusiasm and scaling confidence suggests enterprises may struggle to deliver on AI initiatives without addressing integration and governance issues first.

3. 3

   Trump administration signs orders to accelerate quantum computing development, aiming for a machine capable of scientific research by 2028, as the technology moves closer to commercial use.

   US President Donald Trump signed executive orders to speed up quantum computing development with a target of creating a machine capable of scientific research by 2028. The administration has taken $2 billion（約3200億円） in equity in quantum firms and stakes in other technology companies. Quantum computing is expected to speed drug discovery and materials science, but also poses security risks — Google warned in March that firms should be ready for 'post-quantum cryptography' by 2029. Several quantum firms went public this year, signaling that the previously futuristic technology is moving toward commercialization.

   The 2028 deadline for a quantum machine capable of scientific research, and whether the administration's equity stakes and tech intervention approach will accelerate the timeline compared to private-sector efforts.

4. 4

   I Shot Films for 30 Years. Now I'm Building Safety Systems for AI Agents

   I Shot Films for 30 Years. Now I'm Building Safety Systems for AI Agents

5. 5

   Holden Karnofsky warns that AI safety efforts could have net negative impact, citing governance risks and unintended consequences.

   Karnofsky, a prominent figure in AI safety work, has published a list of downside risks specific to AI safety efforts themselves—acknowledging that even well-intentioned interventions in AI governance could backfire or cause harm. Karnofsky states he takes seriously the possibility that his own impact could ultimately be negative, despite working in what he considers a high-impact cause. This reflects a fundamental tension in safety work: interventions like regulation can easily make things worse, and some could increase the risk of great power conflict.

   Karnofsky emphasizes this is not a comprehensive list, only the risks he personally takes seriously. He frames his work as an effort to act in ways he will be proud of while acknowledging he must live with the possibility of negative utilitarian impact.

6. 6

   Most impactful AI governance work happens invisibly inside government institutions, not in public campaigns—suggesting the policy community may be investing in the wrong visibility.

   A post on LessWrong argues that the majority of strategic AI governance work occurs within ministerial cabinets and international institutions, rather than through visible public channels like press releases and open letters. The AI governance community has historically focused on and celebrated public intellectual work, but the body suggests some of the most consequential decisions are made through invisible insider work in the executive branch. This may indicate a structural mismatch between where resources flow and where outcomes actually happen.

   The author raises concerns about overinvestment in public intellectual production and suggests a bias against invisible types of work. The post specifically questions whether models designed for public visibility—such as replicating ControlAI in France—can succeed in contexts where the real work is hidden from public view.