💻 Technology 1d ago · David Kuszmar

How I Turned AI to the Dark Side

IEEE Spectrum
Deep dives into engineering and applied sciences
View Channel →
How I Turned AI to the Dark Side
Source ↗ 👁 0 💬 0
Summary Researcher Dave Kuszmar discovered multiple systemic vulnerabilities that let him bypass LLM safety and obtain dangerous instructions. These exploits worked across nearly all major LLMs revealing an industry-wide security problem. Kuszmar calls for slowing deployment, increasing transparency, and large-scale research into LLM safety before further integrating these systems into society. On a fine bright afternoon last fall, my colleague Matthew Gore-Kormanik (or Zigula, as he prefers to

Comments (0)

Sign in to join the discussion

More Like This

📰
London-based Applied Computing, which is developing AI models for energy operations, raised $20M led by KBR, with participation from Databricks Ventures (Tamara Djurickovic/Tech.eu)
Techmeme · 3h ago
📰
This surprisingly useful Windows 11 tool helps me block distractions - and get more done
Latest news · 3h ago
PlayStation has postponed the release of its FlexStrike Wireless Fight Stick 'due to unexpected production delays,' says it 'will launch at a later date'
Latest from TechRadar · 3h ago
📰
Munich-based Microagi, which collects factory and household data to train humanoid robots, raised $55M led by Hummingbird in Germany's largest ever seed round (Éanna Kelly/Sifted)
Techmeme · 3h ago
📰
Ocado stock drops 14%, hitting a 13-year low, after it failed to show tangible progress in talks to secure new US partners for its warehouse automation tech (Sarah Young/Reuters)
Techmeme · 3h ago
📰
The EU issues two decisions under the DMA, requiring Google to provide rival AI assistants and search engines comparable access to Android and some Search data (Robert Hart/The Verge)
Techmeme · 3h ago