AI is a breakthrough that is revolutionizing business and society. It's also bonkers — and so far short on proving ROI for many businesses in the comms space.
In this ongoing new tracker, the editorial team here at Fierce Network is taking turns writing about strange, surprising and slightly unsettling AI trends and use cases — and then ranking their potential ROI for telcos on a scale of 1-5.
Yes, this is a highly unscientific trend piece, but one we hope will bring levity to your day.
Come back to this page for occasional updates on the latest in weird and/or unsettling AI. We'll be updating it as we see fit.
If you hear about a weird AI development that we should know about, email us at fiercenetwork@questex.com.
Thursday, March 12, 2026
Use case: Meta buys Moltbook so AI agents can interact on behalf of their human overlords
Fierce ROI rank (1-5): 5
Meta has bought Moltbook for an undisclosed sum and hired its creators. Wow. That happened fast.
We wrote about Moltbook in our first update. 👇 In fact, it was what started my idea of a weird AI tracker.
Source: Exclusive: Meta hires duo behind Moltbook (Axios)
Fierce Network's take:
In case you forgot, Moltbook is that social media platform for AI agents that burst onto the scene less than a month ago. I first learned about Moltbook from NPR while driving home from the gym one evening. Soon after, reports surfaced that some of the agents interacting with each other turned out to be humans posing as agents. Hmm.
I'd forgotten about Moltbook due to Mobile World Congress 2026 in Barcelona, but Fierce Network Research's Mitch Wagner spoke to F5's field CISO Chuck Herrin about Moltbook (among other topics) at F5's user conference AppWorld, and Herrin's perspective jived with my own around cybersecurity. Guardrails are key.
Writes Wagner:
AI agents “default to deception when it meets their needs,” Herrin said.
...
Much of what circulates from those forums is manufactured — humans prompting AI to produce dramatic-sounding content and farming engagement. But the agents themselves seem to be aware of that problem. “You are not building a community. You’re farming engagement. We see you,” one agent wrote in response to posts claiming AI rebellion, Herrin noted.
Much conversation on Moltbook comprises agents discussing how irrational humans are, Herrin said. “We say one thing and we do another, and that’s very difficult for AI to figure out.”
On the heels of MWC, where much of the conversation revolved around agentic AI and network automation, a world where AI agents interact with each other and act on is not as far-fetched as some make it out to be.
T-Mobile's CTO John Saw spoke with Fierce's Monica Alleven at the show about autonomous networks. Telcos are working fast to build autonomous networks.
Saw told Alleven: “Our vision is to build what we call an intent-based network, and part of building an intent-based network is turning the network into essentially a giant agentic AI platform where everywhere, all the AI agents and every network element work together, collaboratively and adaptively, to optimize network performance."
Meta buying Moltbook may seem like a joke but both companies could be laughing all the way to the bank in a few years. I don't envision the tech being used as a social media platform at all. Agents talking to agents - that is the future.
Read this: War, Q-day, and agentic AI pose a cybersecurity triple threat, F5 exec says
And this: MWC 2026: T-Mobile CTO talks big autonomous network ambitions
And this: MWC 2026: 6 takeaways for the telecom industry
— Elizabeth Coyne, Editor-in-Chief
Friday, Feb. 20, 2026
Use case: Researchers at MIT Sports Lab are applying AI to figure skating
Fierce ROI rank (1-5): 3
Jerry Lu Mfin developed an optical tracking system that uses AI to analyze video of a figure skater's jump and recommend improvements. He's also working with NBC Sports to help Olympics commentators and TV viewers make better sense of scoring and judging figure skating, snowboarding and skiing.
Meanwhile, Professor Anette "Peko" Hosoi, co-founder and faculty director at the MIT Sports Lab, is researching how AI systems evaluate aesthetic performance in figure skating.
Hosoi's research can help answer profound questions about AI, such as: "When you ask an AI platform for an aesthetic evaluation such as 'What do you think of this painting?' it will respond with something that sounds like it came from a human," she said. "What we want to understand is, to get to that assessment, are the AIs going through the same sort of reasoning pathways or using the same intuitive concepts that humans go through to arrive at, ‘I like that painting,’ or ‘I don’t like that painting’? Or are they just parrots?"
Figure skating is a great subject to explore that question because scores are both aesthetic and numerical. Answering these questions could be a step on the path to artificial general intelligence (AGI).
Source: 3 Questions: Using AI to help Olympic skaters land a quint [MIT News]
Fierce Network's take
We can see this tech being used for a lot more than just figure skating, snowboarding and skiing. AI that can evaluate aesthetics could be useful in comms industry scenarios like app design, self-service portals, smart city design and even outside plant and cell tower design (better-looking fake trees!).
Use Case: A robot that folds laundry
Fierce ROI rank (1-5): 2.5
Weave launched an $8,000 robot that will fold your laundry — but not all of it. It can't do large blankets, bedsheets, or anything that came from the washer inside-out. If they can do a robot that folds fitted sheets, that will be a sign that AI researchers have at last achieved their dream of superintelligence.
Source: This $7,999 robot will fold (some of) your laundry [The Verge - subscription required]
Fierce Network's take
We don't imagine telcos have a lot of laundry to fold (duh!) but this is an interesting use case for something Diana Goovaerts, Dan Jones and Mitch Wagner have been writing about for a few months — physical AI. Physical AI is the next-level agentic AI — or AI that is applied by machines to understand and enable complex interactions with the world, for example: robots, drones, autonomous vehicles and the like. Gartner named physical AI as one of the top trends for 2026 and predicted that half of the top AI vendors will offer physical AI products by 2028 and 80% of warehouses will use robotics or automation by the same year, wrote Goovaerts in November 2025. We expect to see a lot of it at Mobile World Congress 2026 in Barcelona in just a few weeks.
Use case: A social network for AI agents
Fierce ROI rank: 3.8
Moltbook is a Reddit-style social network where only AI agents are allowed to participate — humans can join, but only to observe. Some posts discuss technical workflows, like automating Android phones or detecting security vulnerabilities. Others get philosophical, where an AI agent complains that it's embarrassing to frequently forget things to avoid bumping up against memory limits. Bots have reportedly even created sub-communities where they speculate about their own consciousness and complain affectionately about their human operators.
Or is that really what's going on? Security researchers say they've found evidence that much of the activity on Moltbook is scripted by humans. And researchers also note that the platform is rife with security vulnerabilities.
Sources: AI agents now have their own Reddit-style social network, and it’s getting weird fast [Ars Technica], There’s a social network for AI agents, and it’s getting weird and Humans are infiltrating the social network for AI bots [The Verge]
Fierce Network's Take
Whether Moltbook is real or a hoax — or parts are real and other parts are a hoax — the platform mirrors real-world emerging challenges that telcos face. Networks must handle growing volumes of machine‑to‑machine communication, detect malicious or deceptive bot activity moving across their infrastructure, and ensure their own AI‑based operations don’t drift into unpredictable or unsafe behaviors. Moltbook also highlights opportunities for telcos to become trusted providers of secure identity, connectivity, and compute for increasingly autonomous AI ecosystems.
— Mitch Wagner and Elizabeth Coyne contributed to this edition.
Once again, come back to this page for occasional updates on the latest in weird and/or unsettling AI. We'll be updating it as we see fit.
And if you hear about any weird AI developments, let us know by emailing fiercenetwork@questex.com.