IT News Digest June 21, 2026: Forced AI Rollbacks, OpenAI's Hires & Relativity Space Mars Deal

The week’s briefing touches on an escalating U.S. intervention around Anthropic’s latest models, the limits of export controls, emerging operational risks from AI agents, strategic hires at OpenAI, a pulled biopic about Sam Altman, and a commercial firm winning a NASA Mars contract.

US action forces Anthropic to withdraw Fable 5 and Mythos 5

The U.S. government required Anthropic to remove its two newest models, Fable 5 and Mythos 5, citing national security concerns. The move followed reports that researchers at Amazon had allegedly discovered a method to bypass Fable 5’s guardrails. Security researchers have protested the government’s intervention via an open letter, and Anthropic has acknowledged related jailbreak issues.

Historical context: export controls haven’t stopped software before

Observers note that attempts to block distribution of sensitive cybersecurity tools haven’t succeeded over the past few decades. Commentators question whether export-control-style measures will meaningfully limit access to models such as Mythos, given prior examples where controls failed to prevent widespread adoption or circulation of security-related code.

Treat AI agents as identities — operational risk and governance gap

Security reporting highlights that AI agents can read data, trigger workflows, deploy code, and interact with core business systems, yet many organizations do not manage them like distinct identities. Token Security and others argue this mismatch creates a growing governance and access-control challenge as agents gain privileges within enterprise environments.

OpenAI hires senior talent ahead of IPO preparations

OpenAI has been recruiting high-profile personnel while gearing up for a public listing, bringing in figures such as Transformer co-inventor Noam Shazeer and former Trump AI policy official Dean Ball. The moves suggest the company is reinforcing technical and policy capabilities as it prepares for the next stage of its corporate strategy.

Amazon MGM drops Guadagnino’s Sam Altman film

Amazon MGM reportedly backed away from Luca Guadagnino’s film about OpenAI CEO Sam Altman, titled Artificial. The project — which cast Andrew Garfield and dramatizes the five-day period in 2023 around Altman’s removal and restoration as CEO — had been in development for roughly a year before the studio dropped it.

Relativity Space picked to fly NASA’s Aeolus payload to Mars

NASA selected Relativity Space, the rocket venture led by Eric Schmidt, to deliver the Aeolus payload to Mars in 2028. Under the public-private partnership, Relativity will supply the spacecraft, the launch vehicle, and cruise operations to carry the payload on the mission.

Taken together, these items underline tensions at the intersection of AI capability, security policy and commercial strategy — from model controls and identity management to staffing moves and big-ticket partnerships in space. Organizations building or governing AI systems should track regulatory actions, tighten agent identity controls, and anticipate how geopolitical and commercial shifts will shape product roadmaps.