eta has open-sourced an AI model that predicts how your brain responds to images, sounds, and speech. The model is called TRIBE v2, and it was built by Meta’s fundamental AI research lab.
It has been trained on over 1,000 hours of fMRI brain scan data from 720 subjects. And it can now simulate neural responses without scanning a single new person.
Hello and welcome back to AI SIX. Six minutes, one story, and everything you need to stay ahead on AI. I’m Joanne Sweeney, and today I want to walk you through what TRIBE v2 actually does, why it matters for anyone working in marketing and communications, and where the governance conversation has to go next.
What TRIBE v2 Actually Does
fMRI stands for functional magnetic resonance imaging. It measures brain activity by tracking changes in blood flow and oxygen levels. It’s expensive, slow, and requires willing participants lying in a scanner for extended periods. Meta’s FAIR lab has spent years collecting over 1,000 hours of fMRI recordings from 720 volunteers who watched films, listened to podcasts, and read text while their brain activity was captured.
That dataset is what trained TRIBE v2. The model processes three types of input simultaneously: video, audio, and text. Each runs through a separate pre-trained Meta AI model. Llama 3.2 handles text. Wav2Vec-BERT handles audio. Video-JEPA-2 handles video. A transformer then brings all three together and outputs a predicted brain map covering approximately 29,000 brain regions across cortical and subcortical areas.
Here is the key capability: TRIBE v2 can make zero-shot predictions. That means it can forecast how a brain will respond to entirely new stimuli, for subjects who were never part of the original study, in different languages, for tasks it has never seen before. It doesn’t need to scan you to model you. In testing, its predictions of group-averaged brain responses were actually more accurate than many individual subjects’ real scans. That’s because individual scans are noisy. TRIBE v2 predicts the statistical average, and it does so with high precision.
For context on the resolution improvement: the original TRIBE model mapped roughly 1,000 brain regions. TRIBE v2 maps approximately 29,000. That’s a 70-fold increase in spatial resolution. The difference is between a rough sketch and a detailed map. And the model’s accuracy continues to scale with more data. There’s no plateau yet.
Where This Is Heading for Marketing and Communications
Right now, TRIBE v2 is a research tool. It’s released under a Creative Commons non-commercial licence. You can download the model weights and run the interactive demo, but you cannot build a product with it or use it commercially. That’s an important distinction, and I’ll come back to it.
But here’s where this is heading. Researchers have already pointed to the neuromarketing implications. A content agency or advertising team could eventually use this kind of model to pre-test how the brain is likely to respond to a video, an audio script, or a piece of copy, before a single person sees it. No focus groups. No A/B testing with live audiences. No lab. Just upload your creative and get back a neural response map showing which moments hold attention, which trigger emotional engagement, and which create cognitive friction.
That is neuromarketing without the neuro lab. And it represents a fundamental shift in how we would validate creative work. We would move from asking ‘did people like this?’ to ‘did their brain respond the way we wanted it to?’ The distinction matters enormously. Because optimising for predicted neural response rather than conscious preference means designing for persuasion at a level that bypasses rational choice.
This isn’t science fiction. The model exists. The capability is documented and reproducible. What doesn’t yet exist is the commercial licence to deploy it at scale, or the governance framework to decide whether we should.
The Governance Question Nobody’s Answering
Meta has been unusually open here. The code, the model weights, the research paper, and an interactive demo are all publicly available. On one reading, that’s responsible open science. It means academic researchers can probe the technology, stress-test its ethics, and publish findings without corporate gatekeeping.
On another reading, it’s a strategic holding position. Meta’s core business is advertising. Understanding how humans respond to content is foundational to that. By making the research open under a non-commercial licence, Meta keeps commercial control in its own hands while the research community debates the implications. That’s open-source in name, but strategically closed where it counts.
And the governance gap is real. If an AI can predict how your brain responds to content you haven’t seen, who decides how that prediction is used? The ethical framework can’t rest with one company. It needs regulators, industry bodies, and the professionals working with this technology to be part of that conversation. Because once predictive neuroscience becomes embedded in content production pipelines, the question isn’t whether to use it. It’s whether the people it’s used on ever agreed to it. And right now, nobody’s asking them.
My Take
TRIBE v2 is a signal, not just a research paper. It tells us that AI is moving beyond analysing what people do and into predicting how their brains respond. The distance between a research preview and a commercial application is shorter than most people realise. We’ve seen that pattern repeatedly in the last three years.
For marketing and communications professionals, the implication is this: the next generation of campaign optimisation may not be built on engagement metrics. It may be built on simulated neural responses. That’s a powerful capability. It’s also one that demands a serious conversation about consent, transparency, and human oversight, before the tools are embedded rather than after.
Know It. Use It. Govern It.
If you want to build your AI literacy and skills, I can help
📚 Grab AI Playbook on Amazon, the perfect podcast companion.
📰 Subscribe to Human in the Loop, my LinkedIn Newsletter
🎓 Take the CPD Accredited AI Playbook Diploma
📞 Book a call at digitaltraining.ie
Key Sources
The Decoder, 27 March 2026 — Meta’s new AI model predicts how your brain reacts to images, sounds, and speech
MarkTechPost, 26 March 2026 — Meta Releases TRIBE v2: A Brain Encoding Model That Predicts fMRI Responses
Meta FAIR GitHub — TRIBE v2 code repository
Neuroscience News — Meta’s TRIBE AI: A New Foundation Model Decoding Human Brain Activity
Digital Applied — Meta TRIBE v2: AI Brain Digital Twins Open-Sourced
Meta FAIR Research Paper — A Foundation Model of Vision, Audition, and Language for In-Silico Neuroscience
Key Statistics
Training dataset: 451.6 hours of fMRI data from 25 subjects
Evaluation dataset: 1,117.7 hours of fMRI data from 720 subjects
Brain regions mapped: approximately 29,000 (20,484 cortical + 8,802 subcortical)
Resolution improvement over TRIBE v1: 70-fold (from ~1,000 to ~29,000 regions)
Zero-shot capability: predicts brain responses for unseen subjects without retraining
Licence: Creative Commons BY-NC (non-commercial)
Released: 26-27 March 2026
Model weights, code, and interactive demo: publicly available
Previous TRIBE v1 won the Algonauts 2025 competition, beating 263 teams
About the Podcast Host
AI and Communications Specialist | Author | Podcaster | International Speaker
Joanne Sweeney is a recognised leader in digital marketing and digital transformation, specialising in the practical application of AI to enhance communication strategies. The founder of Digital Training Institute, Joanne, empowers leaders and communications professionals to leverage AI responsibly and effectively, ensuring transparency, trust, and tangible results.
She completed the Harvard Kennedy School Executive Programme in AI and Public Policy this summer before launching her fourth book AI Playbook, in September. Joanne has also published two editions of Public Sector Marketing Pro, which were preceded by her first title, Social Media Under Investigation Law Enforcement and the Social Web.
Her books have become go-to resources for governments, public sector and policing agencies, as well as marketing and communications professionals, all navigating the challenges of the changing Internet.
Joanne is the host of the AI SIX Podcast, a daily six-minute podcast that provides actionable insights, expert interviews, and practical advice to help professionals navigate the new world of work.
