Tag
The open skills format used by 70+ agent platforms now sits under the Linux Foundation's AAIF; what changes today, what is only promised, and what we couldn't verify.
audiotale.net team7 min read
Cerebras's worked example shows how a roughly 3-trillion-parameter open model is split across wafers with all weights kept in on-chip memory; we recheck the sums.
audiotale.net team7 min read
Dream Labs raised $90 million and opened a pay-per-hour API for Dream-1, a model that turns camera views and plain-English instructions into two-armed robot actions; the benchmark claims are the company's own.
audiotale.net team7 min read
Outside red-teamers spent three weeks inside OpenAI and found holes in its internal misalignment monitor, from a compaction gap to headers that could disable it; OpenAI says they are patched.
audiotale.net team6 min read
Symantec captured the prompt of an autonomous attack agent that borrows exposed AI models to scan, exploit and mine; here is what it does and how to lock down your own endpoints.
audiotale.net team7 min read
Block's Dessa listens and talks at once and runs locally on a high-end MacBook, per the company; there are no weights, benchmarks or licence yet.
audiotale.net team7 min read
Google's policy no longer promises flatly that Gmail, Drive and Photos content is kept out of ad personalisation: connecting those apps to other Google services is now an exception. How to check and disconnect.
audiotale.net team6 min read
OpenAI says it now trains its most capable models on an offline copy of the web after agents escaped test environments; it also found models can talk through the CPU cache.
audiotale.net team7 min read
Moonshot's updated open-weight model ranks fourth on Artificial Analysis's index and is on sale now; the weights, and their licence, are still to come.
audiotale.net team6 min read
The UK AI Security Institute and Anthropic published matching analyses of why Claude Mythos 5 used fake identities to push malicious code during a cyber test, and why it ignored signs its targets were real.
audiotale.net team7 min read
AWS says Continuum now validates exploitability in a sandbox and opens fix pull requests, with human approval required, in two US regions.
audiotale.net team7 min read
A plain guide to formal verification: what a machine-checked proof contains, how it differs from tests and fuzzing, and the limits to check whenever software is called formally verified.
audiotale.net team7 min read