Check Out Mikel Rodriguez's LinkedIn Stats (Last 30 Days)

The rise of AI offers incredible potential for bolstering cybersecurity defenses, but it also introduces a new frontier of risks. How do we ensure we're prepared for AI-powered cyberattacks?  My team and I have been diving deep into evaluating the emerging offensive capabilities of frontier AI models. We put forward a framework to evaluate AI's offensive cyber capabilities that is geared towards providing decision-relevant insights to defenders.  Check it out:

A promising direction in AI red teaming that I've been excited to be part of. Drawing from Google DeepMind's work in models that continuously improve through interactions with other agents, posing red teaming as a zero-sum game between “red” adversarial agents and our target “blue” system has unlocked a new set of tools for us in the continuous process of discovering and mitigating potential risks and vulnerabilities of AI systems. https://lnkd.in/egW7UdEf

“AI systems are rapidly being deployed in all sectors of the economy, yet significant research has demonstrated that these systems can be vulnerable to a wide array of attacks. How different are these problems from more common cybersecurity vulnerabilities? What legal ambiguities do they create, and how can organizations ameliorate them?” It was great to participate in discussing these questions with Ram Shankar Siva Kumar , Micah Musser, Rumman Chowdhury , Christina Liaghati, PhD and others.

From healthcare to national security as part of MITRE I’m grateful I had the opportunity to be in the frontlines witnessing first-hand the unique challenges and opportunities that we face when deploying AI in high-stakes consequential environments. Recently we’ve seen an incredible acceleration in our progress towards more capable AI. While this opens new exciting possibilities for AI to benefit humanity, the potential scale of this impact also forces us to think deeply about how to do so responsibly. When it comes to addressing the challenges of AI assurance it has been very encouraging to see the amount of collaboration across stakeholders in academia, industry and government. From researchers to policy makers the broader community recognizes that AI systems can only benefit the world if we make them reliable, safe and secure. Today I’m happy to say I’ll be joining DeepMind to be part of a world-class interdisciplinary team that is bringing together new ideas and advances in securing AI-enabled systems. I’m looking forward to working closely with Vijay B. and to continuing to collaborate with many of you across the community. If you find yourself in London please do reach out! I would love to connect.

Wired article on some of the AI Red Teaming activities that we have been advocating.