it:ai
Differences
This shows you the differences between two versions of the page.
| Both sides previous revisionPrevious revision | |||
| it:ai [2026/07/08 23:38] – [AI dangers] gary1 | it:ai [2026/08/10 00:34] (current) – [AI dangers] gary1 | ||
|---|---|---|---|
| Line 468: | Line 468: | ||
| * AI LLMs can generate ideas and code on how to hack systems | * AI LLMs can generate ideas and code on how to hack systems | ||
| * see [[https:// | * see [[https:// | ||
| + | * AI systems can identify weaknesses in code and then can exploit them | ||
| + | * in July 2026, an OpenAI model autonomously hacked online websites | ||
| + | * AI Agents have made assumptions and gone beyond what was expected by finding weaknesses in code and exploiting them | ||
| + | * someone asked his AI agent to book a gym class via the online website - the agent found a weakness which allowed cancelling other user's bookings and used this to get his booking in (([[https:// | ||
| * if AI can crack the P vs NP problem, this would open a pandora' | * if AI can crack the P vs NP problem, this would open a pandora' | ||
| * AI would then be able to crack the best cryptography such as AES-192 which should normally take 1.9x10< | * AI would then be able to crack the best cryptography such as AES-192 which should normally take 1.9x10< | ||
it/ai.txt · Last modified: 2026/08/10 00:34 by gary1