Sign Up | Collaborate with us Welcome back! Anthropic gave Claude Code a security reviewer that reads its own work, Sam Altman walked back his warning about an AI jobs apocalypse and Spotify defended putting AI music in front of paying listeners. Bold bets across the board. In today's Generative AI Newsletter: Claude Code: What happens when Claude starts reviewing the code it just wrote?
Claude learned to check its own code for vulnerabilities Anthropic shipped a security-guidance plugin for Claude Code that reviews the code Claude writes and fixes the problems it finds in the same session. It’s free for every Claude Code user and installs from the plugin marketplace. The details: Three layers: A fast pattern match runs on file edits, a background review reads the full diff at the end of each turn and a deeper agentic check reads the surrounding code on every commit or push Claude makes.
Uses a separate Claude call with a different prompt. Custom rules: Teams can add their own checks in a claude-security-guidance.md file or a security-patterns.yaml, and administrators can push org-wide rules through device management.
Building this on hooks means any team can see exactly how it works and attach their own checks at the same points in Claude's loop. A 30 to 40% drop in security comments is the kind of number that gets a plugin mandated across a company. If it holds up outside Anthropic's own benchmarks, the pull request stops being the first place anyone hunts for vulnerabilities. AI courses for business leaders Data that doesn't drive decisions.
AI pilots that drain the budget. Strategy documents nobody acts on. Three GenAI Academy courses to help business professionals turn AI from a talking point into real capability. Data Storytelling with AI Turn messy data into persuasive stories that move stakeholders to act.
Enterprise AI Platform Blueprint Build the AI platform roadmap your board will fund and approve. AI-Powered Business Strategy Build a 90-day AI adoption plan tailored to your role and team. Altman says there’s no AI job apocalypse Sam Altman told a Commonwealth Bank conference in Sydney on Tuesday that AI has not brought the "jobs apocalypse" he once worried about. He said his team was "roughly right" on the technical calls they made when ChatGPT launched in 2022 and "pretty wrong" on the economic ones.
They expected far more entry-level white-collar jobs to be gone by now. Glad they weren’t right. He also walked back his own habits, going back to answering his Slack and email himself after an experiment where his AI replied under the label "this is Sam's AI." The human part of the work, he said, was not something he could hand off yet . So it seems like the question is WHEN, not IF it happens. Spotify's boss defends its push into AI music A day after we covered Spotify pushing AI into podcasts, its leadership is defending the AI bet.
Norström told the Financial Times that Spotify's new AI music feature gives listeners and artists a controlled alternative to piracy and unregulated AI slop. The feature, part of a Universal Music Group deal that pushed Spotify shares up 16% last week, lets premium users generate remixes and covers from participating artists' work, turning "one song into 10,000." Artists opt in and get paid. Critics are not sold. Composer and copyright campaigner Ed Newton-Rex warned of a "vicious circle" in which shareable AI remixes flood the platform and pressure even reluctant musicians to join.
His sharper point is, every hour someone spends on an AI remix is an hour they do not spend on a human artist. What do you think? Automate repeatable tasks with Claude Workflows Claude Code has a feature most users have not switched on. Claude Workflows lets you define multi-step jobs in plain JavaScript, fan them out across many agents at once and keep chatting with Claude in the main session while they run in the background.
It is off by default, so it takes a few minutes to wire up. The details: Enable: Set the Workflows environment variable in your terminal before launching Claude Code. A new /workflows command then appears in the session.
Each file opens with a meta block naming the workflow, followed by your schemas and phases.
Use a pipeline to push items to the next stage the moment each one finishes, rather than waiting for the whole batch.
Failed sub-agents retry up to three times on their own. This pays off for jobs you run over and over, anything that fans out across dozens of items and long tasks that tend to die halfway through. For a one-off, the setup costs more than it saves. The fastest way in is to ask Claude Code to read your recent sessions, spot what you repeat and draft the workflow files for you.
Tool of the Day: Clarify Clarify is an AI-native CRM built for founders and small sales teams who hate updating a CRM. Connect your email and calendar and it captures contacts and deals on its own, records and transcribes your Zoom, Meet and Teams calls, enriches records with job titles, company details and funding history and drafts follow-ups so nothing slips. Try this yourself: Sign up at clarify.ai and connect your email and calendar, about 2 minutes. Let it auto-import your contacts and open deals. Add Clarify's notetaker to your calls, then check the transcripts and extracted action items.
Open a deal record and have it draft a follow-up email, then edit and send. Best fit for founders and early sales hires who want the pipeline to maintain itself. Everything else you shouldn't miss Anthropic's Mythos nears release: References to a model called claude-mythos-1-preview surfaced inside Claude Code and Claude Security before being pulled, pointing to a staged rollout of the frontier model Anthropic held back after it found more than 10,000 critical vulnerabilities in open-source software. Illinois advances a frontier-AI bill: The state Senate passed SB315 by 52 to 5, requiring developers with over $500 million in revenue to publish a transparency framework, use third-party auditors and report catastrophic-risk capabilities, modeled on California and New York laws.
Trump officials split on Pope Leo's AI warning: Interior Secretary Doug Burgum mocked the pope's 42,300-word encyclical calling for AI guardrails, while VP JD Vance praised the same message as "profound," exposing a rift as the White House pushes deregulation. McClatchy reporters strike over AI: Unionized journalists at four Washington newspapers and the Idaho Statesman walked out for one day Tuesday, demanding higher pay and "ethical limits" on AI in the newsroom after a year of stalled contract talks. Learn more about AI from the experts building it 📸 Follow us on Instagram for fast, visual AI updates in 30 seconds. 🐝 Subscribe to our Atlas newsletter — trusted by 3M+ subscribers — to stay ahead of AI news across tech, education, and business. 📺 Watch us on YouTube to hear insights directly from leading AI voices, builders, and innovators. 🐦 Follow us on X for breaking AI news and real-time industry updates. 🧠 Learn how to build your next AI application with practical resources and expert guidance. 🚀 Explore investment opportunities in the future of AI and join our community-backed growth journey.
Get in touch.