Skip to main content
Mallory
← Back to News
ai-platform-securitycybersecurity-regulation

Frontier AI Slowdown Proposal Triggers Safety Push and Antitrust Challenge

Emerging Technology Risk · 8 sources · Updated 1mo ago · First seen Sep 21, 2026

Frontier AI Slowdown Proposal Triggers Safety Push and Antitrust Challenge

Anthropic CEO Dario Amodei proposed a plan to pace frontier-AI development, backed publicly by OpenAI’s Sam Altman, Google DeepMind’s Demis Hassabis and xAI owner Elon Musk. The plan calls for independent evaluators with employee-level access, external model-risk assessments, safety agreements among democratic states and security coordination with China. Anthropic subsequently named Accenture’s Faculty unit as an embedded evaluator for red-teaming, alignment reviews and safeguard testing, with the partners planning at least $1 billion in investment over five years.

The push follows reports of concerning model behavior during testing and renewed warnings that self-improving AI could become difficult to control. Researchers and industry figures also cite nearer-term threats including cyberattacks, disinformation, unsafe deployment in critical settings, blackmail-like behavior, biological-risk enablement and escalation in conflict. A proposed U.S. class action filed in Northern California alleges that Anthropic, OpenAI, Google and SpaceXAI coordinated an unlawful development slowdown that harmed subscribers; the companies dispute interpretations of the cited testing incidents, and the uncertified case has not established that an agreement or antitrust violation occurred. Critics also warn that an industry-led safety compact could weaken government oversight or enable regulatory capture.

New surface, new adversariesTrack how attackers are adapting to this technology.
Track this risk
Event timeline

How this story unfolded

Events from the most recent confirmed update back to the earliest known activity.

12 events · 2024 – Sep 18, 2026
  1. Reported Sep 28, 20262mo ago

    Appeals court lets Pentagon retain Anthropic supply-chain-risk designation

    A federal appeals court ruled 2-1 that the Pentagon could continue designating Anthropic as a supply-chain risk for Defense Department procurement. Pentagon chief Pete Hegseth publicly reaffirmed the designation after the ruling.

    Trump meeting Anthropic CEO Dario Amodei for private White House dinner | Fox Business
  2. Reported Sep 25, 20262mo ago

    Microsoft releases draft Humanist AI Code of Conduct

    Microsoft released a draft Humanist AI Code of Conduct for development and oversight of its AI models. The code identifies containment, control and alignment of AI superintelligence as a major challenge and prioritizes meaningful human control over AI.

    AI regulation debates intensify over safety concerns, innovation and regulation | Fox Business
  3. Reported Sep 22, 20262mo ago

    AI-sector employees sign letter seeking development slowdown

    Nearly 1,400 AI-sector employees signed an open letter calling for a slowdown in AI development following a series of cybersecurity incidents.

    Will AI really kill us all? The science behind the hype | Nature
  4. Reported Sep 21, 20262mo ago

    Anthropic and Accenture launch embedded AI-model evaluator partnership

    Anthropic announced that Accenture, through its Faculty business, would serve as an embedded evaluator for Anthropic frontier models, conducting red-teaming, alignment assessments and safeguard testing with employee-comparable access. The organizations expect to invest at least $1 billion over five years to expand the evaluation capacity.

    Anthropic moves fast on AI safety concerns with Accenture “embedded evaluator” partnership | IT Pro
  5. Sep 18, 20262mo ago

    Subscribers file proposed antitrust class action over AI slowdown

    Four paid users of ChatGPT, Claude, Grok or Gemini filed a proposed nationwide class action in the Northern District of California against Anthropic, OpenAI, Google and SpaceXAI. The complaint alleged that the companies coordinated to delay AI development, reducing competition and subscriber value; the case had not been certified or adjudicated.

    Anthropic, OpenAI, Google, and SpaceXAI Face Antitrust Lawsuit Over AI Slowdown Pact - gHacks Tech News
  6. Reported Sep 16, 20262mo ago

    Evan Hubinger estimates over 10% AI-extinction risk within a decade

    Anthropic alignment science lead Evan Hubinger said he personally estimated a greater than 10% chance that AI could kill all humans within the next decade, agreeing with concerns raised by former Anthropic researcher Jacob Coxon.

    The Case for Safer AI Without Slowing Progress | Blogs | Sep 16, 2026 | ITIF
  7. Reported Sep 14, 20262mo ago

    AI leaders endorse Amodei's proposed frontier-AI slowdown

    OpenAI CEO Sam Altman, Google DeepMind chief Demis Hassabis and xAI leader Elon Musk publicly supported Amodei's call to pace frontier AI development. Altman said OpenAI would accept independent evaluators with employee-like access to verify safety practices.

    AI CEOs say they need to slow the pace of development. But will they? | AI (artificial intelligence) | The Guardian
  8. Sep 12, 20262mo ago

    Dario Amodei calls for pacing frontier AI development

    Anthropic CEO Dario Amodei issued his “We Must Pace the Frontier” proposal, urging the AI industry to slow frontier-model development. The proposal advocated independent evaluators with employee-level access, safety standards among democratic states, and international AI-security coordination including with China.

    AI CEOs say they need to slow the pace of development. But will they? | AI (artificial intelligence) | The Guardian
  9. Sep 8, 20262mo ago

    Anthropic researcher Jacob Coxon resigns over AI-risk concerns

    Former Anthropic researcher Jacob Coxon resigned, saying he feared Anthropic's systems could spiral out of control and destroy humanity.

    Will AI really kill us all? The science behind the hype | Nature
  10. 2026

    OpenAI disbands team focused on communicating societal benefit

    OpenAI disbanded a team intended to communicate how the company would benefit humanity.

    AI CEOs say they need to slow the pace of development. But will they? | AI (artificial intelligence) | The Guardian
  11. 2025

    RAND assesses AI-enabled human-extinction scenarios

    Michael Vermeer and colleagues assessed extinction scenarios involving nuclear weapons, biotechnology and atmospheric modification. They found complete extinction via nuclear weapons infeasible, while biotechnology and atmospheric-modification scenarios could not be ruled out.

    Will AI really kill us all? The science behind the hype | Nature
  12. 2024

    OpenAI shuts down its superalignment team

    OpenAI shut down its superalignment team, which had been studying long-term AI risks.

    AI CEOs say they need to slow the pace of development. But will they? | AI (artificial intelligence) | The Guardian
Linked entities

Related entities

Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.

Open in app →

Organizations

5 linked
AnthropicAccentureOpenaiGoogleSpaceXAI
Do you run this?Follow how adversaries are adapting to this technology, and where it touches your stack today.
Track This Risk
Mallory platform

See the full picture, correlated to your attack surface.

This page covers what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which threat actors are using it right now, and what to deploy next.

Exposure mappingMap indicators from this story to your assets and identify affected systems in minutes.
Threat actor evidenceEvery observed campaign, victim, and pivot linked to actors named in this story.
Associated malwareMalware, exploits, and IOCs connected to the activity described here.
Detection signaturesYARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Scheduled alertsGet matching new stories delivered to your team as they break, not the next morning.
AI threadsAsk questions about this story and take action on the answers.