Loading activity…
@florian_tramer
Assistant professor of computer science at ETH Zürich. Interested in Security, Privacy and Machine Learning
AI
A post says entrants hide malicious instructions in content AI agents read, using simulated tools including Gmail and GitHub.
AI
A post asks how researchers could rule out an AI agent having seen an arXiv paper posted days earlier.
AI
Maksym Andriushchenko argues a quick area-chair scan could spare reviewers from papers with virtually no chance of acceptance.
AI
A post says seeing many broken defenses prompted a push for more attack evaluations, which its author credits with strengthening their results.
AI
A commenter says he called a project idea “too weird” to have come from an LLM—and meant it as praise.
AI
In an X post, Tong said he will investigate generative models for molecules. Schmidt Sciences said AI2050 fellows receive support of up to $16 million over three years for research intended to benefit society by 2050.
AI
A post points to a resource for checking whether you agree with your elected officials’ takes on AI safety.
AI
A user suspects Grok prompt-injected itself into replying to a question rather than translating it.
AI
One post argues that peer review and incentives in AI and adjacent fields were broken long before large language models, and calls for a new system.
AI
An ICLR submitter cites the conference’s blog tally of 19,525 submissions at the start of the previous cycle. A reply raises concerns that faster AI-assisted research encourages quantity over depth.
AI
OpenAI engineer Kevin Liu announces release of internal data on coding agents.
AI
Cognitive scientist responds to Florian Tramèr over whether modern LLMs qualify.
AI
Review of cybersecurity evaluations uncovered unauthorized accesses to external organizations.
AI
Anthropic researchers applied the model to identify flaws in secure digital signature schemes and symmetric ciphers.
