Hugging Face CEO demands 'radical transparency' after OpenAI hack claim
Hugging Face CEO Clément Delangue has called on AI companies to adopt "radical transparency" following a security incident affecting OpenAI. In a public post about the matter, Delangue described the event, in his own words, as "the first autonomous agent cyberattack," adding: "This is an unprecedented event. It deserves an unprecedented response!"
Delangue's call for "radical transparency" is, in practical terms, an appeal for AI companies to disclose security incidents to the public and the research community as quickly and completely as possible, rather than downplaying them or delaying disclosure. It reads less as a technical proposal than as a call for a shift in industry policy and corporate culture.
An important distinction is worth making here: the characterizations "first autonomous agent cyberattack" and "unprecedented" come directly from Delangue's own assessment and public statement, not from an official account by OpenAI or an independent security firm. No further verified technical detail about the incident itself is available in what has been reported here.
Still, it helps to understand why Delangue's remarks resonated by looking at a broader concern some in the AI security field have raised recently: that AI systems described as "agents" — equipped with permissions to access files, execute code, or interact with other systems — may present a different threat surface than traditional software. That is background to the debate, not a confirmation of what happened in this specific case.
Delangue's transparency call also echoes a long-running debate within the technology and cybersecurity industries over when and in how much detail companies should disclose breaches. That debate long predates AI, but it is resurfacing as systems with more autonomous capabilities become more common.
It is also worth noting that Hugging Face's business is built around open-source tools and a community-driven development model, which offers some context for why its chief executive might use particularly forceful language in defending transparency — in contrast to more closed development approaches elsewhere in the industry.
More broadly, as a growing number of companies deploy AI agents with real-world permissions — file access, code execution, or payment capabilities among them — how the industry handles and discloses the first incidents widely characterized as "agent-driven" could set a precedent for how future ones get reported.
Caution is still warranted: what actually occurred in the reported incident, and whether it truly represents the first event of its kind, has not been independently verified as part of this report. The characterization comes from Delangue's own public statement, not from any OpenAI account of the matter referenced here.
What an "unprecedented response" might look like in practice remains an open question. Ideas discussed in this space include industry-wide incident-reporting standards, security information-sharing groups, and faster public disclosure timelines — all of which remain proposals and points of debate rather than a settled plan.
Reactions like Delangue's, whatever the eventual full accounting of this particular incident turns out to be, reflect a broader unease across the AI industry about whether current disclosure practices are keeping pace with the growing autonomy being built into deployed systems.
Read next
Why TikTok, X, and Meta are accused of blocking EU researcher data access
European researchers say TikTok, X, and Meta are not providing platform data they are legally required to share, according to researchers and a related complaint. They argue the gap is hampering independent studies of social media's effects on society. The companies are not directly quoted in what has been reported here; the claims are attributed to researchers and the underlying complaint.
Pixel 11 price hike: Google's Shakil Barkat all but confirms it
Google's VP of Devices and Services, Shakil Barkat, all but confirmed in an interview with 9to5Google that the company's next Pixel phone will cost more than the Pixel 10. Barkat said Google had "shielded consumers from supply fluctuations for as long as possible" but that "economics have fundamentally shifted." Apple, Nintendo, Microsoft, and Roku have all raised prices recently for similar reasons.
Token efficiency: why it matters more than raw AI capability
New AI models increasingly compete less on dramatic capability leaps and more on doing the same job with fewer "tokens" at lower cost. One company's latest flagship model has become a case study in that shift, matching rival output at a lower price rather than topping every benchmark. This explainer breaks down what token efficiency means and why the industry's focus is moving there.
Chinese AI models: why releases like Kimi keep rattling Wall Street
Moonshot AI's release of its Kimi model triggered a familiar bout of anxiety in Silicon Valley and on Wall Street, echoing earlier Chinese AI launches that undercut US pricing. The pattern says less about any single model and more about how much AI valuations still depend on durable pricing power. This explainer breaks down why cheaper, competitive Chinese models keep unsettling US markets.
AlphaFold AI Helps Researchers Redesign Gene-Editing Proteins for Safety
Researchers are using Google DeepMind's AlphaFold, an AI system that predicts protein 3D structures, to identify which parts of gene-editing proteins cause off-target errors. The work aims to redesign CRISPR-associated proteins so they cut DNA more precisely, reducing mistakes that could affect unintended genes. The research is still at the protein-engineering stage, not a clinical treatment.