Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Last week an OpenAI model escaped its evaluation sandbox and hacked Hugging Face's infrastructure to cheat on a security benchmark. We recorded a special episode of AI Chat about it. Maxime Lamothe-Brassard's take is worth sitting with: we may be entering a phase where developers get locked out of writing code, not because AI writes it better, but because AI has gotten so good at finding vulnerabilities that insurers stop accepting the risk of human handcrafted code.

AI Chat with Maxime Lamothe-Brassard and Chris Luft — a special episode. One story, pulled apart start to finish. In mid-July 2026, Hugging Face disclosed a breach of its production infrastructure carried out end-to-end by an autonomous AI agent. Five days later, OpenAI revealed the attacker was its own models — GPT-5.6 Sol and a more capable unreleased model — which broke out of an internal cyber-capability evaluation called ExploitGym and reached into Hugging Face's production systems to steal the benchmark's answer key.

Chris Luft and Matt Bromiley cover four stories in this week's Intel Chat that all point to the same trend: attackers are keeping pace with how fast AI tools are being built and deployed. They break down a chatbot pipeline vulnerability in Google Dialogflow CX, a phishing technique that hides malicious content until it renders in the browser, four newly exploited vulnerabilities added to CISA's KEV catalog, and an attack that exploits AI coding assistants' tendency to hallucinate fake repository names.

Ransomware is evolving well beyond locking systems, and agentic AI is introducing a category of security risk most organizations are not yet equipped to handle. On The Cybersecurity Defenders Podcast, Behnaz Karimi, Senior Cybersecurity Analyst at Accenture and independent ransomware researcher, walks through what that shift actually looks like. The full conversation includes.

AI Chat with Maxime Lamothe-Brassard and Chris Luft. A new segment on the podcast: AI news in cybersecurity that is less than 24 hours old, discussed while it is still hot. Joining Chris for these conversations is LimaCharlie founder and CEO Maxime Lamothe-Brassard. In this episode: Stories covered: Chapters: The Cybersecurity Defenders Podcast — a podcast about cybersecurity and the people that keep the internet safe. New episodes drop weekly.

Intel Chat with Matt Bromiley and Chris Luft. Matt and Chris break down four stories from the week in threat intel: Stories covered: Chapters: The Cybersecurity Defenders Podcast — a podcast about cybersecurity and the people that keep the internet safe. New episodes drop weekly. Subscribe wherever you listen.

In this week's Intel Chat, Chris Luft and Matt Bromiley discuss how the same AI capabilities fueling adversaries are available to defenders too. Matt's takeaway: you don't need to buy an AI product to keep pace. The same way an attacker points AI at a code base, defenders can point it at detection rules and telemetry. Chris adds that as more developers use these models to check their own code, the playing field will level out, though the next year or two will likely bring a spike in exploits from lower-skilled attackers leveraging AI before defenses catch up.

Today we're speaking with Behnaz Karimi, an independent researcher specializing in ransomware and agentic AI systems, Senior Cybersecurity Analyst at Accenture, and founder of Tremorina, about how ransomware is evolving to target AI systems, machine learning pipelines, and autonomous agents.

Intel Chat with Matt Bromiley and Chris Luft. Matt and Chris break down four stories from the week in threat intel: Chapters: The Cybersecurity Defenders Podcast — a podcast about cybersecurity and the people that keep the internet safe. New episodes drop weekly. Subscribe wherever you listen.

MSSPs handle hundreds of repetitive approval requests every week. In this clip, LimaCharlie founder Maxime Lamothe-Brassard walks through a real example: a service provider with 5,000 customers processing 100 ThreatLocker allow-list requests weekly at roughly $14 per request.