OpenAI is reshuffling its safety leadership as the pressure mounts. Johannes Heidecke, head of safety systems, is leaving after two years. He took over in 2024 from Lilian Weng, who now works at Thinking Machines Labs.
Heidecke joined OpenAI in 2021 as an AI safety analyst. His work focused on model alignment, rule-based reward systems, and preparing for dangerous model capabilities. His departure marks another exit among senior safety staff in recent years.
OpenAI’s safety teams have seen turbulence. The Superalignment team, launched in 2023 with 20% of the company’s compute, dissolved in May 2024. The AGI Readiness team folded after leader Miles Brundage resigned in October 2024. The Mission Alignment team lasted just 16 months before disbanding in February 2026.
Meanwhile, OpenAI launched a Safety Fellowship on April 6, 2026, signaling a renewed safety focus. Saachi Jain now serves as interim head of safety systems under VP Mia Glaese. The company insists safety work is integrated closely with model development. Mark Chen, Chief Research Officer, said safety shapes product and launch decisions early on.
The exit of Heidecke follows other notable departures. Fidji Simo, chief of applications, stepped down due to medical recovery. Chief futurist Joshua Achiam left after nine years, claiming on social media there was no specific reason but that the mission can continue outside OpenAI. Achiam recently testified in the Musk v. Altman trial, emphasizing, “The future of humanity depends on the choices we make together about AGI and superintelligence.”
OpenAI also announced product changes. It is merging Codex into the ChatGPT desktop app and adding a new Work setting. The GPT-5.6 model family now offers varied options at different price points. ChatGPT itself nears one billion users.
The company discontinued its Atlas browser in July 2026, shifting browser features into the ChatGPT app. OpenAI describes Atlas’s end not as a failure but a strategic pivot.
External scrutiny intensifies. Forty-two state attorneys general are investigating OpenAI. Meanwhile, Lilian Weng’s new employer, Thinking Machines Labs, voices concerns that AI governance lags behind model capabilities. Founder Mira Murati warns the industry is behind the curve.
OpenAI’s leadership shuffle and product realignments come amid fierce competition and regulatory pressure. The company has also filed confidential paperwork to go public, signaling ambitions beyond research and development.
Despite internal upheaval, OpenAI insists safety remains integral. Mark Chen states safety work is embedded early and directly in model design. Yet with multiple safety teams disbanded and key figures departing, the question remains how deeply safety is woven into OpenAI’s future.
Based on
- OpenAI has folded safety into research again. Its head of safety is leaving. — thenextweb.com
- OpenAI’s chief futurist is leaving the company after nine years. | The Verge — theverge.com
- Apple accuses OpenAI of playing dirty in the AI talent wars | Business Insider Africa — africa.businessinsider.com
- OpenAI’s Browser Isn’t Dead, It Just Moved To The ChatGPT App — engadget.com
- OpenAI is making its biggest play for the office | Business Insider Africa — africa.businessinsider.com
