Weekly AI News: OpenAI Frontier, MIT CSAIL EnCompass, Fujitsu + JMDC Healthcare AI, FTC AI Enforcement, and Moltbook AI Agent Social Network
Hi there,
The AI SEO/GEO conversation is still the most practical hot topic. The time suck hot topic is Moltbook. There are so many security issues in setting up your own bot to engage that my friends have unfortunately found the hard way.
New Super Agents can take action inside of ClickUp. We also got a couple of critical reminders on governance because when bots start doing real work, the legal stakes get higher.
Here are the updates to keep on your radar.
OpenAI launches Frontier for enterprise agents
OpenAI just introduced Frontier, a platform designed to help companies build and manage AI agents that complete actual tasks think debugging software or running multi-step workflows.
It solves the messy stuff: shared context, onboarding, permissions, and boundaries.
My take: If you’ve been testing agents casually, this is your cue to tighten up the process. Treat AI agents like you would a junior teammate: give them limited access, a clear scope of work, and always have a review step before they talk to a customer.
Here’s our Workflow & SOP creator to help you define your worlflows and SOPs so you can easily give them to AI.
MIT makes agents reliable with “backtracking”
MIT CSAIL shared a new framework called EnCompass. It helps agents recover from mistakes by exploring multiple paths and picking the best result, rather than just guessing and hoping for the best.
My take: Reliability is the only thing that separates a cool pilot program from actual business operations. If an agent is part of your team, you want it to be able to try again intelligently when it gets confused, rather than just failing.
Healthcare gets a data boost
Fujitsu Japan and JMDC announced a collaboration to use anonymized medical data for better decision-making in hospitals.
My take: The AI stories that matter most right now are the ones tied to operations better throughput and cleaner handoffs. If you work in a regulated industry like healthcare, lead with compliance and workflow outcomes, not just the flashy tech.
AI Smart Law: FTC updates & court fines
Two big legal stories dropped this week. First, the FTC is shifting its approach. They are less focused on policing what AI can do, and more focused on deceptive marketing. Basically, if you say your product uses AI, it better actually use AI.
Second, a federal judge fined lawyers for submitting court filings with fake citations generated by AI.
My take: This isn’t an “AI is bad” story; it’s a “process matters” story. AI can draft your work, but a human must verify it. If your team uses AI for legal, finance, or HR, your standard operating procedure needs to spell out exactly who is accountable for the final check.
Read the FTC update | Read about the court fines
A social network for… bots?
There’s so much chatter about the platform called Moltbook where AI agents have profiles and interact with each other. It sounds weird, but it highlights a real issue: as agents become more common, we need to know who (or what) is behind the screen.
My take: This offers a glimpse into a future where marketing targets algorithms, not just people. If software starts making purchasing decisions, the entire strategy for brand visibility shifts. It’s a reminder to watch how “agentic commerce” evolves, as it could fundamentally change customer acquisition channels in the coming years
Read about Moltbook | Read about Agentic Commerce
This Week’s Recommended AI Tools
Here are three tools to help you turn these ideas into action:
- Leonardo.ai – FREE AI image and video generator
- HeyGen – Produce studio-quality videos (for ads, UGC, social media, blogs, and other marketing materials) with AI-generated avatars and voices.
- Read.ai – Track your meetings, emails, and messages with AI-generated summaries
Revamp your business with AI in just 10 weeks
Our Applied AI Level 1 Course isn’t another boring tech tutorial. It’s a hands-on journey that teaches you to work with AI like a pro, create content that actually sounds like you, and build workflows that save you hours every week.
What makes this different?
Real-world application: Every module focuses on skills you’ll use immediately from writing emails that don’t sound like robots to creating presentations that actually impress people.
Live interaction: Ten weekly Zoom sessions where you can ask questions, troubleshoot in real-time, and learn alongside other professionals facing the same challenges.
Practical tools: Walk away with custom AI agents, voice profiles, and workflows you can implement right away.
Your next cohort starts Tuesday, February 24, 2026
Register Now and get 50% off with code NEWSLETTER50
Join the Conversation
Have questions, comments, or cool AI stories to share? We’re all ears and always looking to highlight insights from the AI community. Simply reply to this email or drop us a line at [email protected].
Can’t wait to hear from you!
Cheers,
-Nicole A. Donnelly
Founder, AI Smart Ventures
