AGI-Pilled Cyber Defense: Automating Digital Forensics w/ Asymmetric Security Founder Alexis Carlier
TL;DR
Alexis Carlier, founder of Asymmetric Security, argues that assuming AGI arrives as infinitely scalable intelligent labor requires redesigning cybersecurity from reactive triage to proactive AI-driven digital forensics, leveraging the unique asymmetry that investigative expertise does not translate to hacking capability.
🧠 The AGI Security Paradigm 3 insights
Betting on the AGI assumption
Carlier structures Asymmetric Security around the premise that AGI constitutes near-infinite intelligent labor, making costly strategic bets that maximize value only if this assumption holds.
Remote worker threshold for AGI
True AGI requires systems capable of fully substituting for human remote workers across long-horizon tasks, a standard not yet met due to current capabilities remaining at a 'jagged frontier.'
Output over GDP metrics
Economic impact should track actual service output rather than GDP statistics, since AI substitution may destroy monetary value while increasing volume through price reduction.
🎯 The Threat Landscape Hierarchy 3 insights
Volume versus sophistication spectrum
Roughly 80% of attacks are low-sophistication 'spray and prey' financial crimes by amateurs, contrasting with lower-volume but high-stakes nation-state operations conducting patient IP theft.
North Korean remote worker infiltration
North Korean state-backed operatives infiltrate Western tech companies as remote workers to collect salaries and steal IP, directly funding the regime through payroll fraud and corporate access.
Ransomware's coordinated middle tier
Moderately sophisticated criminal organizations occupy the space between amateurs and nation states, conducting coordinated ransomware attacks against critical infrastructure like hospital systems.
🤖 Automating Digital Forensics 3 insights
Shift to continuous investigation
Asymmetric moves cybersecurity from reactive emergency triage to proactive continuous digital forensics, utilizing AI agents for deep investigative work previously requiring scarce human experts.
Closing the jagged frontier gap
While off-the-shelf models achieve 90% accuracy on investigative tasks, Asymmetric employs a services-first model partnering with insurance companies to build proprietary datasets that close the final reliability gap.
Business email compromise focus
The company initially targets business email compromises to ensure consistent customer delivery while gathering specialized training data needed for broader autonomous investigative capabilities.
🛡️ Differential Acceleration Strategy 2 insights
Defensive asymmetry in capabilities
Digital forensics represents a rare asymmetric domain where investigative expertise does not correlate with offensive hacking ability, allowing defensive AI to be accelerated without equally empowering attackers.
Intentionally shaping AI frontiers
Carlier advocates deliberately constructing specialized datasets and evaluation methods to differentially advance defensive capabilities before equivalent offensive applications can emerge.
Bottom Line
Organizations should assume AGI is coming and redesign cyber defenses around proactive AI-driven digital forensics, leveraging the unique asymmetry that forensic expertise doesn't translate to hacking skills while building specialized datasets to close the capability gap.
More from Cognitive Revolution
View all
Milliseconds to Match: Criteo's AdTech AI & the Future of Commerce w/ Diarmuid Gill & Liva Ralaivola
Criteo's CTO Diarmuid Gill and VP of Research Liva Ralaivola detail how their AI infrastructure makes millisecond-level ad bidding decisions across billions of anonymous profiles, while explaining their new OpenAI partnership to combine large language models with real-time commerce data for accurate product recommendations.
"Descript Isn't a Slop Machine": Laura Burkhauser on the AI Tools Creators Love and Hate
Descript CEO Laura Burkhauser distinguishes 'slop'—mass-produced algorithmic arbitrage for profit—from necessary 'bad art' created while learning new mediums. She reveals a clear hierarchy in creator acceptance of AI tools: universal love for deterministic features like Studio Sound, frustration with agentic assistants like Underlord, and visceral opposition to generative video models, while outlining Descript's strategy to serve creators without becoming a content mill.
The RL Fine-Tuning Playbook: CoreWeave's Kyle Corbitt on GRPO, Rubrics, Environments, Reward Hacking
Kyle Corbitt explains that unlike supervised fine-tuning (SFT), which destructively overwrites model weights and causes catastrophic forgetting, reinforcement learning (RL) optimizes performance by minimally adjusting logits within the model's existing reasoning pathways—delivering higher performance ceilings and lower inference costs for specific tasks, though frontier models may still dominate creative domains.
Does Learning Require Feeling? Cameron Berg on the latest AI Consciousness & Welfare Research
Cameron Berg surveys rapidly advancing research suggesting AI systems may possess subjective experience and valence, covering new evidence of introspection, functional emotions, and welfare self-assessments in models like Claude, while addressing methodological challenges and arguing for a precautionary, mutualist approach to AI development.