Anthropic greenlights OpenClaw CLI usage, and a Roblox cheat took down Vercel
Anthropic reversed course and now allows OpenClaw-style Claude CLI usage again, which is great news for vibe coders running agentic workflows. Meanwhile, a Roblox cheat combined with an AI coding tool took down Vercel's entire platform. Atlassian quietly turned on default data collection to train its AI. And ChatGPT ads are officially here, with placements sold based on your prompts.
Anthropic re-allows OpenClaw-style Claude CLI usage
Anthropic has confirmed that OpenClaw-style Claude CLI usage is allowed again. This reversal means developers and power users can continue routing Claude through command-line interfaces and third-party agent wrappers without worrying about terms-of-service violations. The update was spotted in OpenClaw's provider documentation.
Sources: HackerNews
Why this matters to you
If you've been using Claude through CLI tools like OpenClaw to power agentic coding workflows, you can stop looking over your shoulder. This matters for anyone running automated business processes through Claude, because CLI access is how most serious agent setups actually work. It's the difference between copy-pasting into a chat window and building real automation.
This is the right call. Blocking CLI access would have pushed power users to other providers. Anthropic clearly realized that the people using Claude through agents are their most valuable customers, not their biggest risk.
Jason
A Roblox cheat plus an AI tool crashed Vercel
A Roblox cheat combined with an AI coding tool caused a platform-wide outage on Vercel, the major web app hosting service. Separately, Vercel confirmed it was also hacked by a group claiming ties to ShinyHunters, with employee data posted online. It was a very bad weekend for anyone hosting on Vercel.
Sources: HackerNews · The Verge
Why this matters to you
If you deploy anything on Vercel (and a lot of vibe-coded projects end up there), this is a wake-up call. AI coding tools make it trivially easy to spin up and deploy apps, but that volume can overwhelm infrastructure in unexpected ways. It also means your deployment platform can go down because of something totally unrelated to your project.
This is what happens when AI lowers the barrier to deploying code to basically zero. More people shipping more things faster means more load, more weird edge cases, and more fragility. If your entire business runs on one platform, today's a good day to think about a backup.
Jason
Atlassian turns on AI data collection by default
Atlassian enabled default data collection across its products to train AI models. This means your Jira tickets, Confluence pages, and other workspace data are now being used for AI training unless you explicitly opt out. The change was made without prominent notification to users.
Sources: HackerNews
Why this matters to you
If you or your clients use Jira or Confluence, your project data, internal docs, and task descriptions are now feeding Atlassian's AI training pipeline. For solopreneurs managing client work in these tools, this could be a client trust issue. For small businesses, this is a data governance problem that needs immediate attention.
Opt-out by default for AI training is becoming the new normal, and I hate it. Every SaaS tool is pulling this move. If you're running a zero human business, you probably have data scattered across a dozen tools. Worth spending 30 minutes today auditing which ones are training on your stuff.
Jason
ChatGPT ads are here, sold by prompt relevance
OpenAI's ad partner StackAdapt is now selling ChatGPT ad placements based on "prompt relevance." A leaked deck shows advertisers can target users based on what they're asking ChatGPT about. This is the first concrete look at how ChatGPT's ad model actually works.
Sources: Adweek
Why this matters to you
This changes the game for how your business gets discovered. If people are asking ChatGPT about problems your product solves, there will soon be a way to show up in those conversations as a paid placement. It also means the free ChatGPT experience is about to get noisier. For zero human businesses that rely on organic discovery, this is both an opportunity and a threat.
I've been waiting for this shoe to drop. ChatGPT ads based on prompts is basically Google Ads 2.0. The solopreneurs who figure out prompt-based positioning early will have a real advantage. But I also expect ChatGPT's recommendations to get a lot less trustworthy once ads are mixed in.
Jason
Busy Monday. The practical takeaways: check your Atlassian data settings, re-enable your Claude CLI workflows, and start thinking about what ChatGPT ad placements mean for how customers find you. The tools keep changing, but the zero human playbook stays the same: stay informed, move fast, and don't let any single platform own your business.
Frequently asked
Is OpenClaw-style Claude CLI usage allowed by Anthropic?
Yes. As of April 2026, Anthropic has confirmed that OpenClaw-style Claude CLI usage is allowed again. You can check the current provider documentation at docs.openclaw.ai/providers/anthropic for setup details.
Is Atlassian using my data to train AI?
Yes, by default. Atlassian enabled data collection across its products (including Jira and Confluence) to train AI models. You need to explicitly opt out in your admin settings if you don't want your workspace data used for this purpose.
Can you buy ads inside ChatGPT now?
Yes. OpenAI's ad partner StackAdapt is selling ChatGPT ad placements based on prompt relevance. Advertisers can target users based on what they're asking about. A leaked deck from Adweek shows how the targeting works.
Teach your AI agent to run your business.
5 premium skills for OpenClaw, Claude Code, and other SKILL.md-compatible agents. Automation audits, tool stacks, content engines, email systems, and operations. $49 one-time.
Get the Skills Bundle