Take-Two just iced its entire AI division right before GTA 6 — and that tells us a lot

When a publisher kills its entire AI division in the middle of an industry-wide AI gold rush, that’s not just another layoff – that’s a statement about what they actually believe, not what they tell investors.

That’s what’s happening at Take-Two Interactive. The company behind Grand Theft Auto VI has shut down its dedicated AI group and laid off its Head of Artificial Intelligence, Luke Dicken, plus an unknown number of colleagues, just months after loudly talking up generative tech as the future of “efficiency.”

Key takeaways

• Take-Two has closed its central AI division, laying off Luke Dicken and much of his team after seven years of internal tools and research work.
• The move directly contradicts recent public messaging about “actively embracing generative AI” to cut costs and boost productivity.
• For GTA 6, this is a quiet confirmation Rockstar isn’t betting the house on flashy AI systems – it’s still a handcrafted blockbuster first.
• The real story is less “AI is over” and more “AI experiments that don’t ship fast enough get cut just like any other R&D.”

This isn’t just another layoff, it’s a strategic U-turn

The basic facts are simple, even if Take-Two won’t say them out loud. Multiple outlets, citing LinkedIn posts that were later deleted or locked down, report that Head of AI Luke Dicken and an unspecified number of team members have been laid off. European coverage goes further, saying the entire AI division was shuttered.

Either way, at minimum a big chunk of the group is gone. This is the same team Dicken says has been working for roughly seven years on “cutting-edge technology to support game development” – procedural generation, machine learning, generative tools, the unsexy plumbing that makes giant games slightly less impossible to build.

Now place that next to CEO Strauss Zelnick’s recent line to investors that Take-Two is “actively embracing generative AI” to drive efficiencies, right as production on GTA 6 and other blockbusters ramps up. On paper, you don’t take the people driving that strategy and turf them less than 18 months after hiring their leader — unless the plan has changed, fast.

In plain language: Take-Two shuts down its AI division, laying off Luke Dicken and team, at the exact moment you’d expect them to be most useful. That’s either atrocious planning, or a deliberate pivot away from internal AI research toward something cheaper and more short-term.

What that AI team actually did (and what it means for GTA 6)

One thing to clear up immediately: this was not the “NPC brains” or “GTA 6 dialogue” department. This was an internal tech group, spun out of Zynga after Take-Two bought the mobile giant in 2022, focused on tools and workflows across the company’s labels.

Dicken’s own description paints the picture: solve novel problems, build general-purpose systems other teams can plug into, push experimental approaches like generative AI where they make sense. Think content-pipeline automation, asset tagging, smart testing, procedural bits and pieces — the kind of invisible work that doesn’t show up on the back of the box but might shave months off dev schedules if it actually lands.

For GTA 6 specifically, the implication is almost the opposite of what AI panic headlines suggest. Zelnick has already said generative AI can’t build something at GTA 6’s scale and quality on its own, and that Rockstar’s core creative work isn’t getting outsourced to algorithms. Killing a central AI R&D team reinforces that: Rockstar’s next game isn’t going to live or die on some magic AI system.

That doesn’t mean there’s no AI in GTA 6. Of course there is: pathfinding, crowd behavior, police responses, traffic patterns, maybe some machine-learning sprinkled into animation or tuning. But that’s the traditional “AI” game developers have shipped for decades, built by gameplay programmers and tech designers inside Rockstar, not a central corporate skunkworks.

If anything, players worried about AI-written dialogue and generated quests in GTA can take this as an odd kind of reassurance. When you fire the people tasked with proving generative pipelines can massively cut production costs, you’re betting on humans again — or on off-the-shelf tools that don’t demand a dedicated in-house research unit.

The AI gold rush just ran into the realities of game development

Zoom out and this fits a pattern the industry’s been repeating for a decade: new tech arrives, executives promise a revolution on earnings calls, internal R&D teams spin their wheels trying to turn vague hype into shippable features, and then — when the next cost-cutting wave hits — those same teams are “restructured” out of existence.

We’ve seen this cycle with VR, blockchain, the first big wave of live-service tooling, and now AI. The pitch is always the same: this tech will make games cheaper to build, faster to update, and more “engaging” for players. The reality: tooling projects are expensive, hard to integrate into mature pipelines, and rarely deliver the kind of instant ROI finance departments expect.

Look at the rest of the industry. Epic, Embracer, Unity, Sega, Microsoft — everyone has talked up AI-assisted development at the same time as they fire thousands of people to “refocus” on core businesses. The Kinda Funny crew nailed it during the Epic layoffs: even companies with hit products will slash teams when growth curves don’t match the slide deck.

Take-Two is no different. It reports strong results, it has GTA 6 on the horizon, it tells investors it’s embracing generative AI for efficiency — and then it quietly cuts the people who were supposed to build that future. Not because AI is dead, but because the numbers on this particular AI bet didn’t move fast enough.

Is Take-Two suddenly anti‑AI? Not even close

On the surface it looks contradictory: Zelnick publicly dismisses AI’s ability to make a GTA-sized experience, but also tells shareholders AI will make development cheaper, then shuts down the AI team. Underneath, the logic is more boring — and more ruthless.

A central AI division is a long-term bet. It burns money now to (hopefully) save more money later across every label: Rockstar, 2K, Private Division, Zynga. But once the proof-of-concept stage is over, executives start asking awkward questions: why aren’t these tools in production everywhere? Why aren’t milestones shorter? Where are the visible wins we can talk about on the next call?

There are a few obvious ways to answer that pressure:

• Shut down the centralized group and push AI experimentation into individual studios that have immediate needs.
• Stop trying to invent everything in-house and buy external AI services that charge by the API call instead of by the salary.
• Keep using AI quietly for targeted tasks (QA, analytics, pipeline glue) without dedicating a branded “AI division” to it.

Take-Two hasn’t confirmed which of these routes it’s taking — it declined to comment to multiple outlets, which tells its own story. But don’t mistake the death of one internal team for some principled stand against AI. If there’s a cheaper, lower-risk way to sprinkle generative tools into production, the company will use it.

The people actually building games will feel that shift most. Instead of a dedicated partner team designing AI systems around their workflows, they’ll likely be handed third-party tools, corporate “best practices,” or skeleton crews of embedded specialists. The AI doesn’t go away; the support structure around it does.

The uncomfortable part: workers are carrying every AI experiment

The human side here is blunt. Dicken spent roughly a decade at Zynga before the Take-Two acquisition, then less than two years leading a group now being dismantled. His colleagues built careers around a niche that executives have spent the last two years calling “the future,” only to be dropped as soon as that future didn’t instantly pay off.

That whiplash is becoming the default in this industry. Teams are spun up to chase the new thing — live-service economies, NFTs, cloud streaming, AI — and when the story changes or the savings don’t materialize fast enough, those teams are treated as disposable overhead. The tech usually survives in some form. The people who proved it out don’t.

For developers, the lesson is harsh but clear: “AI in games” is not a stable career pillar yet. It’s a layer on top of core skills — engine expertise, online systems, graphics, design — that will absolutely remain valuable. If you’re staking everything on being “the AI person,” moves like this should be a warning about how fragile that niche is when budgets tighten.

For players, the takeaway is simpler. When publishers frame AI as an efficiency tool, what they’re really saying is: “We want the same or bigger games with fewer people and more volatility.” Take-Two’s AI division just found out what happens when that math stops working in PowerPoint.

What to watch next

There are a few clear signposts that will show whether this was a one-off bloodletting or the start of a bigger strategic shift.

• The next Take-Two earnings call: Listen for how often Zelnick still says “AI” and whether he frames it as core strategy or generic “productivity tools.” If AI talk fades, the board’s patience for big bets is gone.
• Rockstar and GTA 6 marketing: If Rockstar starts bragging about AI systems in promo beats (dynamic NPCs, procedural events, etc.), then this work has likely moved into studio-aligned teams. If not, expect a familiar “crafted by humans” narrative instead.
• Job postings across Rockstar, 2K, and Zynga: New roles that mention machine learning, LLMs, or generative tools will show whether AI is being embedded in studios rather than owned centrally.
• Industry-wide AI tooling deals: If Take-Two suddenly pops up as a customer in someone else’s AI press release, that’s the missing piece: the AI work didn’t stop, it just got outsourced.

Those are the metrics that matter more than one company line about “restructuring.” They’ll show how committed Take-Two really is to AI once the hype has worn off and the invoices are due.

TL;DR

Take-Two has shut down its central AI division and laid off Head of AI Luke Dicken and multiple team members, despite recently telling investors it was “actively embracing generative AI.” The move underlines that GTA 6 and other flagship games are still being built as traditional, hand-crafted blockbusters, with AI staying mostly in the background as tooling, not as headline features. The practical read: AI isn’t going away at Take-Two, but long-term internal experiments just lost to short-term cost-cutting — and the people who built them are the ones paying for that decision.