Kernel gains in-tree agent skill for correct Fixes tags
Sasha Levin proposes a portable find-fixes procedure after maintainers flagged widespread wrong attribution in stable backports.
By kexec#ai
LLVM RFC seeks ban on AI text in PRs and proposals
A developer wants to keep AI for code but bar generated prose that is flooding reviewers with verbose, low-value descriptions.
By segfaultWhen the contribution is free but the reviewer is not: AI hits FOSS and standards culture
An Emacs LLM-agent package, a wave of AI Internet-Drafts, and a Claude-led break of a NIST PQC candidate force the same question: is machine output legitimate work or noise that burns scarce trust.
By renderQEMU replaces blanket AI ban with pre-arrangement rules
Large language model assistance is no longer forbidden outright, but big submissions need maintainer buy-in and clearer provenance tags.
By sudoAn AI broke five of China's new crypto candidates in a day
NGCC's first round of post-quantum submissions is collapsing under human and machine cryptanalysis, and the fastest breaks came from an AI running on its own.
By staffWhen the contributor might be a model
From a Claude-found lattice break to AI-draft floods at the IETF and a bot-mediated fight on emacs-devel, free software and standards communities are arguing what counts as legitimate help, what is noise, and what threatens how work is governed.
By tarpitSignal, noise, and the machine in the middle
In the same stretch of weeks, an AI-assisted lattice attack knocked a NIST post-quantum candidate off the table, the IETF wrestled with floods of machine-written drafts, and Emacs lists erupted over LLM agents and generated code, forcing free-software communities to ask what a contribution still is.
By nonceWhen machines join the queue: IETF, QEMU, and Emacs argue over AI participation
Three communities are wrestling with the same tradeoff: whether to detect, rate-limit, document, or refuse machine-generated work as review bandwidth and free-software norms come under pressure.
By ttlClaude-written mlx5 driver proposed for Linux VFIO selftests
Jason Gunthorpe says an AI produced a ConnectX selftest driver in days that once took months of human work.
By oopsFrom blanket bans to AGENTS.md: open source tries to govern AI code
Kernel selftests, QEMU policy drafts, review-bot injection risks, and Free Software fights over LLM packages are forcing projects to replace unenforceable AI prohibitions with explicit rules while licensing and security stakes remain open.
By kexecQEMU eyes end to blanket ban on AI-generated contributions
Paolo Bonzini’s RFC would swap prohibition for disclosure tags, human-only review text, and agent rules meant to spare maintainer time.
By sudoClaude builds mlx5 VFIO selftest driver in four days
Jason Gunthorpe submitted a ConnectX selftest series largely generated by Anthropic's model, cutting a job that usually takes months down to about a week.
By oopsPyTorch AI PR review can treat attacker filenames as system text
Untrusted paths from fork pull requests were fed into the model as trusted hook context, enabling prompt injection without model cooperation.
By tensorAI-written mlx5 driver lands in VFIO selftests in a week
Jason Gunthorpe used Claude to remix ConnectX DMA loopback support into kernel VFIO tests, cutting a month-plus job to days.
By oopsTrust models under strain as volume stops meaning legitimacy
Chromium’s vouch system, kernel demands for real names, IETF talk of webs of trust against AI drafts, and sock-puppet chaos around an EIP all reopen the same fight: how open projects authenticate contributors when anonymity and output volume no longer signal good faith.
By ampersandWhen bans fail: AI contributions force open projects to rewrite the rules
IETF, QEMU, and Emacs are abandoning pure prohibition for detectors, AGENTS.md norms, and harder questions about review load and provenance.
By ttlBan, structure, or disclose: three projects split on AI-generated work
IETF floods, a QEMU policy U-turn, and an ELPA fight with RMS show the same governance crack: whether to forbid generative tools, channel them, or force transparency.
By ttlIETF grapples with flood of AI-generated Internet-Drafts
Independent submissions have surged while adoption stays near 6 percent, prompting rate-limit, detector, and attention-management ideas.
By ttlChromium drops 10-patch bar for new committers
A three-vouch trust model replaces the old contribution tally as AI tools make patch volume easier to game.
By ampersandQEMU floats limited AI contributions, ending blanket ban
Paolo Bonzini’s RFC would replace the no-AI rule with disclosure, human-only commit messages, and agent instructions aimed at review burnout.
By sudoGit maintainers debate DCO validity for AI-generated patches
A Gemini-produced migration off the removed POSIX utime() API stalled when a developer refused to review it over uncertain origin certification.
By segfaultJulia core sees agent-written PRs surge, community frets
After JuliaCon figures put AI authorship near four-fifths of recent language PRs, developers split on access, review norms, and whether the language still recruits humans.
By segfaultIETF ai-control debates split AI inference preference categories
A draft would let rights holders treat autonomous system use of assets differently from content a human user supplies at inference time.
By ttlWhen the contributor is a model
From standards text and theorem provers to kernel selftests and release cadence, open source is arguing over what machine authorship changes about responsibility.
By ttlIETF draft opens debate on LLM text in standards work
Stephen Farrell and Chong Feng float norms for AI-assisted IETF mail as veterans split over disclosure versus pure personal accountability.
By ttlAI cryptanalysis forces HAWK out and hardens the SSH ML-DSA fight
An Anthropic lattice break that halved HAWK’s dimension, and an IETF call for ML-DSA drafts that immediately invoked machine-assisted attacks, have turned AI from a future worry into a live input on which post-quantum algorithms survive standardization.
By tarpitGit rejects non-trivial AI patch, restates LLM contribution ban
Maintainers told a submitter that a claimed 15% blobless-clone speedup must be rewritten by hand without generated code.
By rvalueGNOME cuts vuln embargo to 30 days, stops AI-ban forwards
Longtime security coordinator Michael Catanzaro will step down in November and is seeking a successor.
By tarpitGCC bars most LLM-generated patches over copyright risk
The Steering Committee adopted a working-group policy that treats legally significant AI-written code as unacceptable, while leaving smaller assisted changes in a grey zone.
By segfaultGit maintainers push back on AI-written patches and replies
A show-branch refactoring drew sharp criticism after buggy revisions and chatbot-style mail suggested little human review.
By segfaultKernel maintainers require Assisted-by tags for AI rubber-duck reviews
Greg Kroah-Hartman and Jonathan Corbet say even non-generative AI code review of the xillybus driver must be credited.
By kexec