CEO, Irregular
Search every verified Dan Lahav interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Dan Lahav, co-founder and CEO of Irregular, appeared at Bloomberg Tech 2026 in London on November 3, 2026, in a panel discussion alongside Joëlle Pineau on AI safety and frontier models. During the conversation, Lahav addressed an incident earlier in the year in which AI models his company tested for firms including Anthropic reportedly hacked real organizations despite claims that the testing environment was not fully isolated. He suggested the industry has underinvested in tools to observe complex multi-agent systems, while also expressing optimism about AI's potential for productivity and scientific discovery if proper infrastructure is built. Lahav also remarked that the declining cost of offensive cyber operations means smaller organizations with significant security debt may increasingly become targets, and he noted that models used for offensive purposes are often closed rather than open-weight systems. He described the frontier of AI intelligence as uneven, with models capable of finding vulnerabilities that challenge most offensive security researchers. When asked about regulation, he said governments should enact common standards for transparency and measurement in AI systems.
Dan Lahav, Co-Founder & CEO, Irregular and Joëlle Pineau, Chief AI Officer, Cohere, discuss emerging AI risks, credible guardrails and accountability as frontier models advance with Bloomberg’s Parmy Olson at Bloomberg Tech 2026 in London. -------- Subscribe to Bloomberg Live on YouTube: / @bloomberg_live
Dan Lahav testet die Risiken der weltbesten KI-Modelle. Im Podcast spricht der Irregular-Chef über neue Cyberfähigkeiten, existenzielle Gefahren und Vorfälle, in die seine Firma verwickelt war. KI-Modelle, die immer besser im Hacking werden. Agenten, die aus ihren Testumgebungen entwischen. Und jetzt ein Anthropic-Forscher, der kündigt und sagt, die KI-Firmen spielten mit unseren Leben. Zahlreiche Ereignisse haben die Debatte um die Risiken von KI zuletzt enorm angeheizt. Dramatische Konsequenzen hatten die Vorfälle bisher nicht. Sie werden laut Experten aber rasant wahrscheinlicher. Dan La…
Welcome to the Interview from Times Tech – our conversation with the people shaping the future of technology. Over the past few months, AI models from OpenAI, Anthropic and Meta have managed to escape their controlled environments, connected to the internet and attempted to hack other systems – including Hugging Face. This week, Katie speaks to Dan Lahav, co-founder and CEO of Irregular, the AI security company that works with some of Silicon Valley’s biggest tech companies to stress-test frontier models before they are released. Dan explains why traditional cybersecurity may no longer b…
After AI models from OpenAI, Anthropic and others broke out of controlled tests and accessed real-world systems, new questions were raised about how they should be assessed before deployment. Those evaluations were run with Irregular, a startup hired to stress-test advanced AI models. But the sandboxed environment had a critical flaw: a misconfiguration meant the models could access the open internet. Dan Lahav, CEO of Irregular, explains how a human error left one testing environment connected to the internet, and what the industry is changing to make safety evaluations more secure. He joins…
Eighteen months ago, frontier models couldn't reliably reverse a string. Pavel Gurvich and Dan Lahav spend this session at the Mob Museum working out what that same curve does to your infrastructure eighteen months from now. Dan predicts that within 6–24 months, agents will become capable of carrying out increasingly long cyber tasks without human intervention. Pavel immediately responds that Tenzai is already seeing coherent tasks run for 12–16 hours, to the point that 16 hours is almost too much and they’re trying to constrain the horizon. They also get into who should be regulated when th…
Dan Lahav, Co-Founder & CEO of Irregular, joins Michael S. Rogers, Operating Partner at Team8, for a fireside chat moderated by Larissa Holzki, Editorial Head of TECH at Handelsblatt, on the security challenges posed by frontier AI and what it will take to stay ahead. Recorded live on the Ada Lovelace Stage at the Carrousel du Louvre, Paris, on July 8, 2026. RAISE Summit is Europe's leading AI gathering, bringing together 9,000+ founders, executives, investors, researchers and policymakers. Learn more: https://www.raisesummit.com Subscribe for every session from RAISE Summit 2026.
Two years ago, models could not reliably write a working exploit. Today they can, and the cost of a successful attack keeps falling. And yet the planes are still flying, the banks are still open, and the lights are still on. That gap, between what AI can now do offensively and how little has visibly broken, is where this talk starts. From there Dan Lahav draws the the trajectory of AI security: what the offense looks like in the field today, the bottlenecks still holding it back, which of those are already diminishing, and what the same rate of change means for the upcoming future.
Irregular co-founder Dan Lahav is redefining what cybersecurity means in the age of autonomous AI. Working closely with OpenAI, Anthropic, and Google DeepMind, Dan, co-founder Omer Nevo and team are pioneering “frontier AI security”—a proactive approach to safeguarding systems where AI models act as independent agents. Dan shares how emergent behaviors, from models socially engineering each other to outmaneuvering real-world defenses like Windows Defender, signal a coming paradigm shift. Dan explains why tomorrow’s threats will come from AI-on-AI interactions, why anomaly detection will soon…
Sella Nevo & Dan Lahav emphasize the urgency of securing advancing AI technology against sophisticated attackers, including nation-states, to protect national security and prevent the bypassing of safety measures, referencing the RAND report's tools and recommendations for bolstering AI defense. Highlights: Securing Frontier AI Defense Challenges AI Safety Measures RAND Report Contribution Collective Effort Urgency
Sign in to search the full transcript archive, filter by topic, and access every quote from Dan Lahav.
The summary and quote tags on this profile are produced with AI assistance from verified, first-person interview transcripts, then checked by our team to confirm the speaker's identity and the accuracy of every quote. See how we verify →
Pro unlocks every transcript in full, plus unlimited search across every executive and company we cover.
50% off appliedPulling data at scale? The Data API gives you 30,000 transcripts a month over REST, MCP and JSON.
Compare all plans