Insights · April 1st, 2026
Here we share Nicholas Carlini – a Research Scientist at Anthropic – speaking at the [un]prompted 2026 conference on: Black-hat LLMs and how they uncover vulnerabilities (vulns).
Large language models are now capable of automating attacks that were previously only possible by human adversaries. In this talk, I discuss several ways that adversaries could mis-use current models in order to cause harm both at a larger scale and at a lower cost than they do currently. For example, we find that recent state-of-the-art models can now find 0-day vulnerabilities in large software projects that have been extensively tested by humans for decades. These new capabilities will alter the threat landscape and require we rethink security in the coming years.
What this means for CEOs
- The threat has fundamentally changed in scale and cost – The core message isn’t just “AI is a security risk” — it’s that the economics of cyberattacks have collapsed. Adversaries can now execute large-scale, cost-effective attacks that were previously only possible by skilled human hackers. That means the barrier to entry for sophisticated attacks is gone. Threat actors don’t need elite talent anymore; they need access to a model.
- Your legacy software is newly exposed – Claude found a blind SQL injection in Ghost’s Content API — a flaw that allowed a completely unauthenticated user to compromise the admin database and take full control — in 90 minutes. Ghost had never had a critical severity vulnerability in its entire history. If a pristine, well-maintained codebase can fall that fast, CEOs need to assume that older, less-scrutinized internal systems are at significant risk. When pointed at some of the most well-tested codebases — projects that have had fuzzers running against them for years, accumulating millions of hours of CPU time — the model found high-severity vulnerabilities, some that had gone undetected for decades.
- The attack surface is everything, not just crown jewels – In a post-attention-scarcity world, successful exploit developers won’t carefully pick where to aim. They’ll aim at everything — operating systems, databases, routers, printers. These kinds of targets run everywhere, including in every regional bank and hospital chain in North America. The assumption that attackers will prioritize your most valuable assets no longer holds. Automated tools will sweep broadly.
- The vulnerability pipeline is industrialized – Carlini described running a trivial bash script across every source file in a repository, prompting the model to find exploitable vulnerabilities, then running a second pass to verify exploitability — with a near-100% success rate. This is not sophisticated hacking. It’s a commodity workflow. Any motivated actor can replicate it.
- Patch cycles are now dangerously slow by comparison – Chrome’s vulnerability submissions in March 2026 already exceeded twice the total from February, and Firefox saw a similar surge — with roughly 25% of all bugs found in an entire prior year being submitted in a single batch. Your IT and security teams are not staffed for this volume.
What CEOs should act on now
- Use AI defensively before adversaries use it offensively against you. The most promising use cases include autonomous security auditing for projects that can’t afford dedicated red teams, accelerated patch verification, and triage prioritization for organizations drowning in CVE noise.
- Rethink your security vendor relationships. Traditional penetration testing cycles (annual, quarterly) are structurally obsolete. You need continuous, AI-assisted vulnerability scanning.
- Pressure your software supply chain. These weak points — routers, printers, networked components — were priced into everyone’s cost of doing business. To patch them, someone has to physically intervene. Know what you’re running, especially on-premise infrastructure that doesn’t auto-update.
- This is a board-level conversation. The threat model CEOs were briefed on 18 months ago is already outdated. The capability curve here, as Carlini noted, doubles roughly every four months. What’s possible today will seem primitive by year end.
Now is the time to shore up your cyber capabilities and give them the ability to use LLMs (carefully with consideration) to review and profile your systems. And, we need to ask our software vendors what they are doing about this as well
About Nikolas Badminton
Nikolas Badminton is the Chief Futurist & Hope Engineer at futurist.com. He’s a world-renowned futurist keynote speaker, consultant, author, media producer, and executive advisor that has spoken to, and worked with, over 500 of the world’s most impactful organizations and governments.
Nikolas is an artificial intelligence expert and his 2026 keynote ‘The AI Leader: Create Incredible Productivity, Profit & Growth’ is the level up for the modern CEO and executive leader.
Please contact futurist speaker and consultant Nikolas Badminton to discuss your engagement.
