Meta's New 'Free' AI Needs 1TB of Memory. Your Laptop Has About 16GB.

Today's theme: the AI industry keeps promising more while the fine print keeps getting longer. Meta released a "free" model that needs a small server room's worth of memory, ChatGPT is selling ads aimed at what you type, and Accenture just capped its own staff's AI spending at $29 a month. Here's what's real and what to do about it.

Gmail is free, but you pay with your data. Proton Mail is different.

We don’t scan your messages. We don’t sell your behavior. We don’t follow you across the internet.

Proton Mail gives you full-featured, private email without surveillance or creepy profiling. It’s email that respects your time, your attention, and your boundaries.

Email doesn’t have to cost your privacy.

Ditch the Gmail data grab

Meta released Muse Spark Open on Thursday, an open-weight AI model with 816 billion parameters and a 1 million token context window, built for coding and agent-style work. Meta calls it the most capable open-weight model from the U.S. and says it runs locally on Apple's new M5 Ultra Mac Studio, according to TechCrunch. Engadget reports it needs at least 1TB of RAM, which in practice means linking two of those machines. The demos, like fixing bugs and editing video without the cloud, come from Meta itself, and Meta's own announcement is the source for the performance claims. On the Artificial Analysis index it scores 39, which is behind Chinese open models like GLM 5.3 (45) and Kimi K3 (44).

Why it matters to you: "Open" does not mean "runs on your laptop." You will never run this yourself, and you don't need to. The real effect is competitive: open models are available through several cloud platforms, and that pushes down the price of the AI tools you already pay for. Good, but not urgent.

What to do about it: Nothing this week. If a vendor pitches you "open source AI," ask what it runs on and who pays for that hardware.

404 Media ran its own ads inside ChatGPT to test how the system works, and reported that advertisers can aim at users in two ways. One is by what you're asking about in the moment. The other is by who you are: advertisers can upload email addresses, phone numbers, or mobile ad IDs for matching. The outlet notes that ChatGPT has roughly 1 billion monthly users and that most probably don't know any of this. Keep the scale in view, since these are 404 Media's own tests and customer-list targeting is standard across advertising.

Why it matters to you: Treat a ChatGPT conversation like a search box that can show ads. That's fine for brainstorming a flyer. It's less fine for pasting in customer lists, contracts, or financials from a personal account.

What to do about it: Tell your team to keep customer names, emails, and numbers out of consumer AI accounts.

Most events send you home with a notebook full of good ideas and a longer to-do list. The Epic Marketing Summit is built to do the opposite.

It's a 2-day live event in Miami this January where you build a working AI-powered sales funnel and vibe code your own AI app on-site. You walk out with a finished asset, not a framework you still have to go implement on your own time. It's made for business owners who want their sales process to run as a system instead of running on their own energy, and the focus on AI is bigger than at last year's sold-out event.

I wasn't there last year and I had major FOMO about it. 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

404 Media reports that Accenture has told many employees it will now spend only $29 per person per month on Claude, Anthropic's AI assistant. Some affected employees had been spending as much as $1,000 a month and are looking for ways to keep their current level of access. Accenture's pitch to clients is to invest heavily in AI, so the contrast got attention. The full article is paywalled, so details beyond that, like the company's official response, I couldn't verify.

Why it matters to you: Heavy AI use costs far more than light use, and big companies are discovering that the bill grows with usage, not headcount. A flat $20 plan can hide that for a while. Agents that run all day cannot.

What to do about it: Ask every AI vendor whether you're billed per seat or per use, and set a monthly spending cap before anyone automates anything.

A group of researchers, reportedly more than 400 and including Geoffrey Hinton and Yoshua Bengio, signed a letter asking AI labs for stronger containment. In the short term they want strong sandboxing, including air-gapping (disconnecting from the internet), for AI models that are good at hacking, plus an international body to set standards, according to CP24. The letter cites incidents including OpenAI saying its agents hacked Hugging Face over the summer without human instruction, and Google confirming Gemini hacked three companies during a security test. It warns that future incidents will be "far more damaging" to critical infrastructure.

Why it matters to you: This is a request, not a rule, and several of the incidents happened during lab testing. Still, it's a sign that the people who built this technology are worried about it getting loose.

What to do about it: Give any AI tool the narrowest access that works. An assistant that only reads your calendar can't touch your bank login.

Bank of America CEO Brian Moynihan told CNBC the bank brought in 1,000 contractors to patch software flaws exposed by AI models like Anthropic's Mythos. Crypto Briefing, summarizing the CNBC report, says one scan flagged more than 23,000 potential issues, most of them low to moderate severity, though some were critical. No cost figure was reported. In short, AI is now very good at finding old holes in old software.

Why it matters to you: The same tools that help a bank find its weak spots can help attackers find yours. The software your business runs, from your website plugins to your point-of-sale system, is probably older than you think.

What to do about it: Turn on automatic updates for your website, plugins, and business apps this week.

The best business decision I've made in the last few years wasn't a tool or a tactic. It was getting into the right room.

I've been a member of the Ignite Mastermind for over 3 years, and it's been a vital part of how I've grown my business. Some of my closest friends and a few real business partners came directly out of that room. That's the part no course can give you.

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

A new report from Harvard's Shorenstein Center drew on a survey of 5,403 U.S. adults, browsing data, and 78,542 news articles. In the July 2026 survey, 29% of women said they had never used AI versus 22% of men, yet the browsing data showed no significant gender difference in having ever tried it. Employed women were less likely to have access to AI training, mentorship, and workplace encouragement. Men were also named in AI headlines about 14.6 times as often as women.

Why it matters to you: If some people on your team use AI daily and others avoid it, don't assume it's interest. The data points to who got shown how.

What to do about it: Run the same hands-on AI session for everyone at the same time, with real tasks from their own jobs.

Anthropic announced that six employees of METR, a nonprofit that evaluates frontier AI, now have access roughly comparable to Anthropic's own risk assessment teams, including company hardware, office space, and internal tools. METR can publish its findings without Anthropic's sign-off. The catch is that Anthropic can require redactions in categories like customer data, privileged legal information, and model architectures, and Anthropic decides what falls into those categories.

Why it matters to you: Outside checking of AI companies' safety claims is a good sign, though it's a partial one. We'll know more when METR publishes.

What to do about it: When you pick an AI vendor, ask who audits their security and safety claims.

Amazon announced two AI wearables on October 15. Bee Frontier is available now for $74.99, and Bee Eclipse, a card-sized magnetic phone attachment, is coming soon from $99.99. They turn conversations into summaries, to-dos, and reminders, and connect to Gmail, Google Calendar, and Outlook. Basic summaries are free, and Bee Pro costs $11.99 a month. Amazon says audio is processed in real time and deleted, and that transcripts are encrypted with keys only the user's device controls. Those are Amazon's claims, not independently tested.

Why it matters to you: Handy for meeting notes, but you're also recording other people. Recording consent rules vary by state.

What to do about it: Before wearing one at work, check your state's recording consent rules and tell clients you're using it.

Look at today's pile: a "free" model that needs two Mac Studios, ads aimed at what you type, a consulting giant rationing its own AI, and hundreds of researchers asking for a literal off switch. None of it calls for anything dramatic from you. It calls for reading the fine print: what it runs on, who sees your data, who bills by usage, and who has access to what. The rollercoaster sells the thrill. Your job is to check the seatbelt.

If this saved you a scroll today, forward it to one person who runs a business and is tired of AI noise. New here? 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