Newsom Pushes AI 'Kill Switch' as States Fill a Federal Vacuum
California Governor Gavin Newsom signed an executive order on September 18 accelerating independent oversight of frontier AI systems, including third-party audits, safety plans from outside verifiers, and groundwork for an AI "kill switch" [4][5][6]. The order builds on existing state laws (SB 813, AB 1405) and tasks experts with delivering recommendations — including updated definitions of AI safety incidents — within two months [5].
Newsom framed the move as a response to both the unchecked risks of advanced AI and what he characterized as federal inaction on meaningful regulation [4][6]. Supporters argue California is stepping into a genuine governance gap, arguing that waiting for Washington could mean waiting too long given the pace of AI development.
Critics counter that the order amounts to state overreach — an unelected panel of experts effectively deciding what counts as "good" or "bad" AI, with potential chilling effects on innovation [4]. The debate echoes broader fights over tech regulation: is proactive caution prudent stewardship, or premature interference in a fast-moving industry that doesn't yet know its own risks?
AI Giants Sued Over Alleged 'Pacing' Cartel
A federal antitrust lawsuit filed September 18 in the Northern District of California accuses Anthropic, OpenAI, Google, and SpaceXAI of illegal collusion under the Sherman Act [7][8][9]. The suit, brought by consumers including attorney Cheyenne Hunt, points to public calls for coordinated "pacing" of AI development — most notably Anthropic CEO Dario Amodei's essay on slowing the frontier for safety reasons, which was publicly endorsed by figures including Elon Musk, Sam Altman, and Demis Hassabis [8][9].
Plaintiffs argue this amounts to a self-serving cartel that sidesteps binding public regulation while limiting competition and consumer choice, dressed up in the language of safety [7]. The companies' defenders counter that voluntary coordination on existential-risk questions is exactly the kind of responsible behavior critics of AI often demand — and that treating safety collaboration as antitrust violation could discourage exactly the caution regulators like Newsom are separately trying to mandate [9].
The case puts a sharp point on an unresolved question in AI governance: when does industry self-regulation become anti-competitive behavior, and when is it a necessary stopgap while formal rules catch up?
The Bigger Picture
Today's stories share a common thread: institutions — media, government, and industry — are all wrestling with who gets to define the rules of engagement, and who gets to police them. Trump's press ban raises questions about the boundaries of executive power versus press independence; Newsom's AI order asks whether state governments should preempt federal regulators; and the antitrust suit against AI labs questions whether voluntary safety pacts are principled leadership or quiet collusion. In each case, reasonable people land in very different places depending on which risk they weigh more heavily — abuse of power or paralysis through inaction.
What makes these disputes valuable rather than merely divisive is that they resist easy resolution. A journalist can genuinely believe both that biased coverage exists and that government retaliation against the press is dangerous. A technologist can champion AI safety while still worrying that "pacing" agreements entrench incumbents. Holding these tensions simultaneously — rather than collapsing them into a single tribal position — is precisely the kind of disagreement worth having out loud.
These are not stories with tidy villains and heroes; they are stories about competing legitimate concerns colliding in real time. The measure of a healthy public conversation isn't whether we agree on the outcome, but whether we can articulate the strongest version of the other side's argument before rejecting it.
Key takeaway: Power, safety, and accountability are colliding across media, government, and tech today — and the most productive response isn't picking a side reflexively, but understanding why thoughtful people land on different ones.
Sources
- https://www.thewrap.com/media-platforms/politics/trump-bans-cnn-ms-now-politico-white-house/
- https://www.politico.com/news/2026/09/18/trump-white-house-media-ban-politico-01084438
- https://www.nytimes.com/2026/09/18/business/media/trump-cnn-ms-now-politico-white-house-ban.html
- https://www.gov.ca.gov/2026/09/18/governor-newsom-issues-executive-order-to-accelerate-independent-oversight-and-advance-the-creation-of-an-ai-kill-switch/
- https://www.nytimes.com/2026/09/18/technology/ai-safety-california-gavin-newsom.html
- https://www.politico.com/news/2026/09/18/newsom-california-executive-order-ai-01083826
- https://www.politico.com/news/2026/09/18/anthropic-openai-spacexai-google-sued-over-calls-to-pace-ai-development-01085023
- https://news.bloomberglaw.com/litigation/openai-anthropic-google-spacexai-hit-with-antitrust-lawsuit
- https://www.unite.ai/consumers-sue-anthropic-openai-spacexai-and-google-over-alleged-ai-pact/