AI #170: Lack of Executive Order

2023-08-29 · Source: Don't Worry About the Vase · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Cybersecurity & Data Privacy · Depth: Advanced, extended

Summary

The week saw the indefinite postponement of a US federal Executive Order on AI, reportedly due to lobbying efforts, leaving American AI policy ad hoc. Concurrently, Illinois passed SB 315, a state-level AI safety law requiring third-party auditing, similar to California's SB 53 and New York's RAISE Act. The *Magnifica Humanitas* encyclical was revealed, emphasizing common good over AGI or existential risk, though its practical interpretation is debated. Other significant developments include Google DeepMind's AlphaProof Nexus solving nine open Erdos problems, Anthropic's Project Glasswing identifying over 10,000 cybersecurity vulnerabilities, and the increasing prevalence of sophisticated AI-generated content winning awards and challenging human discernment in writing and art. Microsoft also cancelled internal Claude Code licenses due to cost, shifting developers to GitHub Copilot CLI.

Key takeaway

For Directors of AI/ML navigating rapid technological shifts, prioritize robust internal AI governance and invest in tools to detect sophisticated AI-generated content. The cancellation of federal AI orders and rise of state-level regulations like Illinois SB 315 underscore a fragmented policy landscape, demanding proactive compliance and strategic planning to mitigate risks from evolving AI capabilities and widespread AI-driven misinformation.

Key insights

AI's rapid, broad advancement challenges regulation, human discernment, and societal structures.

Principles

• AI regulation is complex and often ad hoc, with state-level initiatives emerging.
• AI-generated content is increasingly sophisticated, making human-AI distinction difficult.
• Advanced AI models demonstrate novel problem-solving capabilities in mathematics and cybersecurity.

Method

Leverage AI to identify and automate low-hanging workflows, or use a meta-agent approach where one AI directs another to handle tasks and corrections.

In practice

• Use LLMs like GPT-5.5 Pro or Learned Hand for legal motion distillation.
• Employ AI systems like Anthropic's Mythos Preview for large-scale vulnerability detection.
• Utilize AI for personalized health dashboards by integrating DNA, blood test, and wearable data.

Topics

• AI Regulation
• AI Capabilities
• AI-Generated Content
• Cybersecurity
• Large Language Models
• AI Ethics
• Compute Infrastructure

Best for: CTO, VP of Engineering/Data, AI Architect, Policy Maker, Director of AI/ML, Tech Journalist

Related on AIssential

• Trump loses more control over AI regulation as Illinois passes landmark law — Illinois's new AI safety law mandates third-party testing and incident reporting for frontier models, setting a national regulatory precedent.
• Cyber Lack of Security and AI Governance — Advanced AI models like Mythos are rapidly enhancing cyber capabilities, forcing governments to confront regulatory challenges and inter-agency power struggles.
• 😼 AI's Biggest Problem Just Changed. Nobody's Ready. — AI's primary challenge has shifted from capability to governance and control over autonomous agents.
• 😸 The TL;DR on the White House's new AI plan — The White House seeks to centralize AI regulation federally, preempting state laws and defining key policy areas.
• Trump Postpones AI Order Because of Concerns About Overregulation — President Trump postponed an AI oversight order, fearing regulation would hinder U.S. competitiveness against China in the global AI race.
• Trump abruptly cancels EO signing event after top AI firm CEOs declined to go — The US faces internal and external pressures balancing AI innovation with government-mandated safety testing, impacting its global regulatory standing.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Don't Worry About the Vase.