Hacker-friendly AI is changing everything
Notas del episodio
The latest AI models, which have serious cybersecurity capabilities, are having a rapid impact on policy in the US, where the Trump administration has suddenly realised it’s not a great idea to maintain its hands-off policy to AI development. But in Europe, which lacks early access to models like Anthropic’s Mythos and OpenAI’s GPT-5.5, the situation presents a threat to which the European Commission has no useful answer.
Tech Gets Real is a show about staying in control in a time of rapid technological change. I'm veteran tech journalist David Meyer, and I've been covering topics like online privacy and freedom for two decades. If you like this, please hit like and subscribe, and share it with others who might find it useful.
REFERENCES:
UK AI Security Institute on GPT-5.5 https://www.aisi.gov.uk/blog/our-evaluation-of-openais-gpt-5-5-cyber-capabilities
My ISMG article on the MEPs’ letter to the Commission https://www.govinfosecurity.com/european-meps-push-for-stronger-post-mythos-cybersecurity-a-31599
Wednesday’s European Parliament IMCO committee debate
CNN on OpenAI wanting to expand GPT-5.5 access https://edition.cnn.com/2026/04/29/tech/openai-cybersecurity
New York Times on Trump administration’s change of heart https://www.nytimes.com/2026/05/04/technology/trump-ai-models.html
US Center for AI Standards and Innovation announces early access deal with Google, Microsoft and xAI https://www.nist.gov/news-events/news/2026/05/caisi-signs-agreements-regarding-frontier-ai-national-security-testing