Import AI 452: Scaling laws for cyberwar; rising tides of AI automation; and a puzzle over gDP forecasting

2025-10-13 · Source: Import AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, Robotics & Autonomous Systems · Depth: Intermediate, long

Summary

Lyptus Research has identified a clear trend of increasingly advanced AI models performing more sophisticated cyberattacks, with a doubling time of 9.8 months across frontier models since 2019, steepening to 5.7 months since 2024. Models like GPT-5.3 Codex and Opus 4.6 achieve 50% success on tasks taking human experts 3.1-3.2 hours. Open-weight models like GLM-5 lag by 5.7 months, suggesting rapid diffusion of offensive capabilities. Separately, INSEAD and Harvard Business School research shows startups adopting AI for internal use significantly outperform peers, with treated firms discovering 44% more AI use cases, completing 12% more tasks, and generating 1.9x higher revenue. MIT research indicates a "rising tide" of AI automation, projecting 80-95% success rates for most text-based tasks by 2029. However, a Forecasting Research Institute study reveals a paradox: experts expect rapid AI progress but only minor GDP growth (around 1% by 2030).

Key takeaway

For CTOs and VPs of Engineering assessing strategic technology investments, prioritize deep integration of AI across your organization's core functions, especially product development and operational efficiency. The data strongly suggests that early and sophisticated AI adoption is a critical differentiator for competitive advantage and capital efficiency, leading to significantly higher revenue and task completion. Simultaneously, be acutely aware of the rapidly advancing offensive cyber capabilities of AI, necessitating robust defensive strategies.

Key insights

Advanced AI models are rapidly enhancing cyberattack capabilities and driving significant performance gains for startups.

Principles

• AI capabilities expand broadly and consistently.
• Internal AI adoption boosts startup performance.
• Offensive AI capabilities diffuse rapidly.

Method

Lyptus Research evaluated AI models on cybersecurity benchmarks and a new 291-task dataset. INSEAD/HBS conducted a field experiment with 515 startups, providing AI integration workshops to a treated group.

In practice

• Integrate AI across product development and strategy.
• Automate accounts receivable with AI.
• Use AI for bootstrapping and improving startup margins.

Topics

• AI Cyberoffense
• AI Scaling Laws
• Startup AI Adoption
• AI Automation Impact
• Economic AI Forecasting

Code references

• lyptus-research/cyber-task-horizons-data

Best for: CTO, VP of Engineering/Data, Executive, AI Scientist, Director of AI/ML, Policy Maker

Related on AIssential

• Import AI 452: Scaling laws for cyberwar; rising tides of AI automation; and a puzzle over gDP forecasting — Advanced AI models are rapidly improving in both offensive cyber capabilities and business integration, driving efficiency and economic shifts.
• this is really bad... — AI significantly amplifies cyberattack capabilities, necessitating advanced AI-driven defenses and highlighting geopolitical risks.
• The AI Competitive Map Through the Scaling Paradigms — AI scaling is accelerating, with differentiation shifting from raw capability to mastery of distinct scaling paradigms.
• The gap is widening — Autonomous AI agents represent a new paradigm, but their emergent complexity creates significant enterprise adoption and risk management challenges.
• Everyone's getting hacked — AI is rapidly accelerating both cyber attack capabilities and defensive measures, creating an "AI vs. AI" security paradigm.
• The Sequence Radar #816: Last Week in AI: $110B Bets, Nano Banana 2, and the New Economic Reality — AI is rapidly reshaping economic structures and geopolitical dynamics through massive investment and advanced model capabilities.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Import AI.