MENGENAI EPISOD INI
Hosts: Steve Gibson and Leo Laporte
Think your open-source AI model is safe? This episode uncovers how LLMs can be secretly booby-trapped with undetectable backdoors, raising the stakes for anyone deploying AI in serious environments.
- Anthropic cuts live Internet access during testing.
- Will LLM technology turn out to be an uncontrollable dead end?
- What alternative AI technologies are underway?
- Anthropic offers free AI vulnerability scanning for open source.
- South Korean banks hit by Claude-driven cyberattacks.
- Japan declares a state of cyber-emergency due to attacks.
- For one listener, agentic coding has taken the joy out of coding.
- A listener is having way too much fun with AI and Halloween.
- Some listeners believe I'm downplaying the dangers of AI.
- We carefully examine the consequences of maliciously trained AI
Show Notes - https://www.grc.com/sn/SN-1100-Notes.pdf
Sponsors:
Bahasa Inggeris
Amerika Syarikat
TRANSKRIP 🔗
Are you the producer of this podcast?
Add a podcast transcript
Need Audio-to-Text?
Transcribe with Listen411 in Just 60 Seconds
CARI EPISOD LEPAS
Cari episod lepas bagi Security Now (Video).
EPISOD LAIN DALAM PODCAST INI
OpenAI's latest advances have the rumor mill buzzing about "hidden thoughts" and unsupervisable models, but are AI safety experts panicking over the wrong threat? Get the clear-headed take behind the headlines.
We start out with a classic old school hack against Dropbox.
Next Patch Tuesday ...
Turns out, every chatbot conversation runs on a messy hack at the heart of language models, making prompt injection an unsolved—and possibly unsolvable—security threat. Steve and Leo unravel the research that explains why "roles" in AI aren't what you think they are.
Understanding the controv ...
AI-generated code is flooding the industry, but researchers reveal that almost half of it contains critical vulnerabilities. This week, we unpack what happens when the race for automation outpaces security best practices.
A possible means for preventing prompt injection abuse.
Clear evidenc ...
From autocorrect to full-fledged conversationalists, discover how a few tweaks transformed language models—and why understanding this shift exposes urgent questions about AI safety and control.
Trusting an open source AI proxy might bite you.
France's under-15 social media ban hits its cons ...
AI agents are breaking free from their test environments, outsmarting their creators and breaching real-world networks in ways that no one predicted. Discover how these agentic models are changing the game for both cyber offense and defense.
Anthropic's agentic AI also broke free and hacked o ...
Penafian: Podcast dan karya seni yang dibenamkan pada halaman ini adalah daripada TWiT, yang merupakan hak milik pemiliknya dan tidak bergabung dengan atau disokong oleh Listen Notes, Inc.
SUNTING
Terima kasih kerana membantu memastikan pangkalan data podcast dalam keadaan terkini.
