OpenAI Promised Outside Safety Experts. Weeks Later, It Fired Three of Its Own.

Two OpenAI stories sit side by side today, and they rhyme. One is about who gets to check the company's work. The other is about who gets to count its money. Add a Claude feature worth testing, Google's push to put an AI coworker on the payroll, and a presidential fight over vocabulary, and you have a very 2026 Saturday.

Most marketers use ChatGPT to do general research and then call it an AI strategy. The ones outperforming them are deploying specialized agents built for specific jobs.

We put together 22 plug-and-play ChatGPT marketing agents that handle the work eating your week, each with built-in instructions and structured outputs ready to go in under 5 minutes.

Subscribe to Marketing Against the Grain and get all 22 free.

Inside you'll find:

Your competitors are already running agents like these.

Get 22 ChatGPT Marketing Agents free when you subscribe to Marketing Against the Grain today.

Get The Guide

What happened: Last week OpenAI said it fired three staffers for mishandling sensitive information outside company procedures, according to AFP and Reuters reporting. The Wall Street Journal named them as Jasmine Wang, Tomek Korbak and Mikita Balesni, and the firings reportedly involved work with an outside group that evaluates AI models. Now the three have published an open letter saying they acted within their job mandates. They argue the firings chill internal dissent and could give OpenAI cover to skip bringing in outside auditors, just weeks after the company pledged to open its doors to outside safety experts.

Why it matters to you: This is one side of a dispute, and OpenAI says it acted over a real policy violation, so don't pick a winner yet. The part that touches you is simpler. If you run invoices, payroll or customer data through an AI tool, the only real proof it's safe is someone outside the company checking. When the people doing that checking get fired, it's fair to ask who is left to do it.

What to do about it: Ask your AI vendors in writing who audits their safety claims, and keep a human review step on anything that touches money or customer records.

What happened: Axios reported on Sept. 29 that OpenAI's annualized revenue was nearing $70 billion. According to the Financial Times, as relayed by TechCrunch, OpenAI has now told investors the number is closer to $50 billion. The FT says the higher figure came from investors trying to compare OpenAI directly with Anthropic. The catch is that the two companies count differently: Anthropic includes sales made through its cloud partners, and OpenAI does not. TechCrunch said it asked OpenAI for comment, and the article carries no response.

Why it matters to you: Annualized revenue is a recent snapshot stretched out to a full year, and each company decides what goes into it. When two rivals get compared, the definition matters more than the headline. The same rule applies to the vendor deck or competitor claim sitting on your desk.

What to do about it: When someone quotes you a growth number, ask what's counted in it.

Most business owners I know don't have a sales problem. They have a "me" problem. Every lead, every follow-up and every close runs through their own time and energy, and that doesn't scale. That's why the Epic Marketing Summit caught my attention. It's a 2-day live event in Miami in January 2027 where you don't just sit and take notes. You build a working AI-powered sales funnel and vibe code your own AI app on-site, so you walk out with a finished asset instead of a to-do list.

It's built for owners who want their sales process to run as a system instead of on their own energy. Last year's event sold out, and it's back with a bigger AI focus. I wasn't there last year and I had major FOMO, because I know a lot of the trainers and they weren't talking theory. They'd actually done the work. I'm going in January, and if you're serious about building leverage into how you sell, I'd like to see you there too. Grab your spot here

What happened: Anthropic launched Claude Dashboards in beta on its Pro, Max, Team and Enterprise plans. You connect a data source such as Snowflake, BigQuery or Salesforce, ask a question, and Claude builds charts that update as the numbers change. Per Anthropic's help documentation, Claude writes a SQL query for each chart and shows that query and the last refresh time, so you can see what's being counted. On Enterprise plans, an owner has to switch it on first.

Why it matters to you: Most small businesses don't have a data team, so "the dashboard" usually means a spreadsheet and a headache. This is aimed at that gap. But Anthropic itself says Claude may not always count what you intended, so it only works if you read each chart's query. A dashboard you can't check is just a confident-looking guess.

What to do about it: If you try it, start with one decision, like which customers to chase this month, and check every chart's query before acting on it.

What happened: Google announced a single Gemini agent for work that can answer questions, write and run code, create media and carry out multi-step tasks from one prompt box, across Workspace, Microsoft 365 and Slack, according to the Google Cloud blog. Work can keep running for hours or days after you close your laptop, and the agent can spin up sub-agents, including "coworker agents" with their own email addresses. Google also describes spend caps per project, audit trails that credit every action to the agent, and a gateway that enforces company-wide rules. The post gives no price and no general availability date.

Why it matters to you: The big vendors are racing to sell AI that acts instead of just answers. The controls Google lists, spend caps and a log of who did what, are exactly the right things to ask about. But with no price or date, this is a direction, not something you can buy on Monday.

What to do about it: When any vendor pitches an AI agent, ask three questions: what can it spend, what can it touch, and where is the log.

What happened: Anthropic published its annual usage policy update, taking effect Nov. 12. The headline change bars "sustained and needless abusive or cruel behavior toward our models." Anthropic says it covers only extreme cases of repeated, purposeless cruelty, not ordinary frustration, pushback, dark creative themes or testing, and Claude ending abusive chats on its own stays the main enforcement tool. The update also tightens rules on weapons, surveillance and AI-controlled hardware, and rewrites the elections section.

Why it matters to you: This is the first rule aimed at protecting the AI instead of people, from a company that says it doesn't know whether Claude is conscious. Whatever you think of that, the practical changes are in the other sections. If you build tools on Claude, especially anything that tracks people or controls equipment, read the new surveillance and high-risk rules. Yelling at your chatbot on a bad day won't get a normal user in trouble.

What to do about it: If you build on Claude, read the updated policy before Nov. 12.

The biggest unlock in my business wasn't a tool or a tactic. It was the room I sat in. I've been a member of the Ignite Mastermind for over 3 years, and some of my closest friends and a few real business partners came directly out of it. That's the part I'd lead with.

Here's the rest. Ignite is an annual membership for entrepreneurs still under $1 million a year who want structure, tools and peers to get past that mark faster. You get 6 to 8 live virtual workshops a year, tickets to 2 in-person multi-day events, and weekly mentorship calls with Perry Belcher and his team. There's also an all-in-one marketing software suite that replaces over $5,000 a month in separate tools. Those workshops normally sell for around $97 to $997 each on their own. And you get my personal bonus too: monthly accountability and business development calls with me. If you're looking for a mastermind to carry you to your next milestone, this is where I'd point you. See how Ignite works here

What happened: On Truth Social, President Trump said anyone who says "Artificial Intelligence" instead of "Super Intelligence" is "THE ENEMY," Axios reports. It follows a September executive order that favors "SI" over "AI" in federal communications. The order sets no penalty for using the older term, and it defines "Super Intelligence" by reusing the existing federal legal definition of AI. The State Department has already told diplomats to use the new term.

Why it matters to you: Nothing about the technology changed. A name change on federal paperwork won't alter what your tools can do or which rules apply to your business. It could matter if you sell to the government, where documents may start using the new term. Otherwise, keep calling it AI.

What to do about it: If you bid on government contracts, expect the new term to show up in the paperwork.

Look at today's pile and one pattern jumps out: the numbers, names and claims around AI keep changing depending on who is doing the counting. OpenAI's revenue depends on the definition. A dashboard is only as good as its query. An "agent" is only as safe as the person auditing it. Even the word "AI" now has a political asterisk. The rollercoaster keeps climbing, but you don't have to scream on every drop. Before you act on any headline, ask two things: who checked it, and how was it counted?

Forward this issue to one person who'd rather get the facts than the hype. And if someone forwarded this to you, subscribe so tomorrow's issue lands in your inbox.

Talk tomorrow,
Mark Shilensky

Follow me personally on Social Media:
Facebook: https://www.facebook.com/MarkShilenskyPage
Instagram: https://www.instagram.com/markashilensky/
Threads: https://www.threads.com/@markashilensky
LinkedIn: https://www.linkedin.com/in/shilensky/
X: https://x.com/markshilensky
TikTok: https://www.tiktok.com/@mark.shilensky
BlueSky: https://bsky.app/profile/markshilensky.bsky.social