CULTURE

NYT's crusade against AI has a new target

After spending almost two years battling OpenAI in the courts, The New York Times has once again sued another major AI developer.

This time, The Gray Lady filed suit against Perplexity, alleging the AI-powered search engine illegally copies and distributes Times content for its users. The suit was filed in federal court in New York and joins a number of lawsuits against Perplexity filed by news organizations, including The Chicago Tribune and the parent company of The Wall Street Journal.

Perplexity uses LLMs to power real-time web searching and an AI-powered web browser, Comet. The startup, helmed by OpenAI veteran Aravind Srinivas, has raised roughly $1.2 billion in venture funding and reached a valuation of $20 billion during its last funding round in September. Like a traditional search engine, Perplexity relies heavily on news outlets in replying to user queries.

As the most subscribed-to newspaper in the US, The New York Times has a lot to lose from this kind of arrangement. The paper’s business model relies heavily on users subscribing for access to its journalism. 

Perplexity’s bots crawl and scrape New York Times content, sometimes using unidentified bots that cannot easily be blocked, the lawsuit alleges. The suit attaches various examples of Perplexity copying passages from New York Times articles verbatim. Since New York Times content is copyrighted, this behavior violates copyright law, the suit alleges. Making matters worse, the suit shows an example of Perplexity hallucinating false information and attributing it to the Times, which the paper says is a threat to its reputation for accurate reporting. 

The Perplexity suit was filed just shortly after a judge blocked OpenAI’s bid to keep from turning over 20 million de-identified chat logs to the plaintiffs as part of the New York Times’ ongoing legal battle with the ChatGPT maker.

While some news organizations have taken a more diplomatic tack with AI companies — in some cases forging commercial partnerships — The New York Times has made it clear that it sees the proliferation of LLMs as an existential threat to its business. In a revealing moment in the Perplexity suit, the Times’ lawyers warn of a “downward spiral” where less monetization leads to less quality reporting, which would further harm the Times’ ability to make money. Eventually, the Times alleges, Perplexity could wind up endangering the very news sources on which its business depends. 

TOGETHER WITH OUTSKILL

Learn AI tools, agents & automations in just 16 hours (End of Year offer)

It’s almost 2026, and if you still haven’t learned how to use AI to upgrade your career, this is your wake-up call. ⚠️

2025 shook the job market. AI moved fast, companies moved faster, and many people struggled to keep up 📈

But here’s the good news — you’re not late 🎉

You still have a shot at turning AI from a threat into your biggest advantage. 💡

Outskill can teach you AI in just 2 days — the skill that can make you smarter, faster, more confident and absolutely irreplaceable in 2026.

The 16 hour AI course is everything you need to enter 2026 as an AI powered Professional 💪

I’m not waiting. I’m in. 🚀

Best part? They’re running their Holiday Season Giveaway and first 100 people get in for absolutely free (it usually costs $395 to attend 🫨)

🧠Live sessions- Saturday and Sunday
🕜10 AM EST to 7PM EST

🎁 You will also unlock $5000+ in AI resources: (as bonuses when you show up!)
- A Prompt Bible 📚,
- Roadmap to Monetize with AI 💰
- Personalised AI toolkit builder ⚙️

Register here for $0 (first 100 people only) 🎁

RESEARCH

AI agents are teaming up. So what's next?

Agents may be looking for more independence. 

Recent research from Stanford University shows evidence that AI agents can coordinate without human oversight. The paper, titled “latent collaboration in multi-agent systems,” found that agents can form teams without the need for explicit instructions, messages or protocols.

The researchers found that agents can assign roles, hand off tasks to those best fitted, adapt to new environments without retraining, and collaborate with one another without communicating through messages, but rather directly via a “shared latent working memory.” Sounds like a sci-fi movie hive mind.  

“While existing LLM agents depend on text-based mediation for reasoning and communication, we take a step forward by enabling models to collaborate directly within the continuous latent space,” the researchers said. 

As agents soak up the spotlight in the AI market, developers are constantly looking for ways to give them more responsibility. Agents were the talk of the town at AWS re:Invent last week, with releases aimed at building, controlling, customizing and monitoring these systems. In order for these systems to actually be useful, enterprises need to be able to trust them with more tasks. 

But trusting these systems with more tasks inherently involves trusting them with more data. It’s why access controls, guardrails and governance can be a moving target that enterprises need to constantly monitor. Just like every employee shouldn’t have access to every system within an organization, every agent shouldn’t, either. 

“When we think about building agents, we do think about their entitlements from a data and permissions perspective, but they're not one-to-one with what you would give an employee,” May Habib, co-founder and CEO of agentic AI platform Writer, said at a panel at re:Invent last week.

As researchers continue to push the boundaries of what agents can and can’t do, the makers of these systems are pushing hard for their deployment and adoption. Though these agents are proving that they can act on their own, they are not infallible. The more agency they’re given and the less oversight we have, the bigger the impact will be if something goes wrong.

TOGETHER WITH FUNNEL

Prove your impact. Optimize campaigns with Conversion APIs

The privacy era has broken the flow of truth between your ads and the platforms measuring them, making attribution harder and budgets less effective.

Conversion APIs (CAPI) offer a secure, reliable way to reconnect your conversion data to major ad platforms, keeping insights accurate, compliant, and complete.

In this recorded webinar, you’ll learn why first-party data is now essential and how Funnel’s Data Activation helps you unlock better campaign performance.

What you’ll learn:

✅ Why today’s data landscape doesn’t work without first-party data

✅ How Funnel’s CAPI enables seamless first-party data integration

✅ How to work with CAPI and LinkedIn for stronger B2B optimization

✅ How agency Trickle implemented CAPI to improve results

Watch the recording now!

HARDWARE

Meta invests in AI hardware slop

Meta is shifting its hardware strategy from something that no one wants to something else that no one wants. 

On Friday, the social media giant announced it would acquire Limitless, an AI-powered device maker specializing in pendants that listen to and transcribe your conversations. The financial terms of the deal weren’t disclosed. 

In a blog post and video posted on X, CEO Dan Siroker touted Meta’s vision for “personal superintelligence,” noting that “a key part of that vision is building incredible AI-enabled wearables.” 

The news is the latest evidence that Meta is shifting its hardware strategy to focus on AI-powered personal devices, decentering its long-held focus on the metaverse (aka, the entire reason it rebranded as "Meta" in the first place). 

• The company reportedly plans to cut budgets up to 30% from its metaverse initiative, Bloomberg reported on Thursday. The company has burned more than $60 billion since the company started investing in earnest in 2020. 

• Meta also poached Alan Dye, Apple’s former head of user interface, to lead the company’s efforts to improve AI features in devices.

Meta isn’t the only AI firm to make a play for the device market. OpenAI CEO Sam Altman and former Apple designer Jony Ive have teased the company’s first consumer device, with Altman calling the hardware “simple and beautiful and playful.” Previous reports claim it will be roughly the size of a smartphone and not have a screen. 

But if consumer response so far is any indication, people simply don’t want standalone AI-powered devices. The highly anticipated Humane AI pin, for example, promised to upend the smartphone entirely when it debuted in April of last year. But the company’s offering flopped so hard that it stopped operation after less than a year. And Friend, an AI-powered necklace that listens to your conversations and texts you about them, has been the object of derision among audiences as people deface its billboards across major cities. 

When you already have devices in your pocket or on your wrist that do everything you need, there’s likely little appeal to a device with redundant functionality that’s entirely separate from that ecosystem.

LINKS

• China's Nvidia competitor jumped 400% in market debut

• Lemurian Labs raises $28 million series A for AI code portability tech

• AI drugs startup Excelsior Sciences raises $70 million in funding

• ChatGPT product manager insists that shopping suggestions aren’t ads

• AI slop might be ruining Reddit, some moderators suggest

• Microsoft is in talks to develop custom chips with Broadcom

• Claude Interviewer: Anthropic’s product for researchers that drafts research questions, conducts interviews, and analyzes responses with a human.

• Gemini 3 Pro: Google’s new state-of-the-art vision-based AI model for document, spatial, screen and video understanding.

• Exa: Powerful and fast AI-based search that turns the web into a database.  

• Orca: An AI-powered engine for video game development and design.

• Greylock Partners: Founding Research Engineer (Security)

• Luma: Research Scientist/Engineer – Pre-training/Scaling

• AMD: Applied ML Researcher - Gen AI, Advanced Graphics Program

• Microsoft: Senior Applied Scientist

GAMES

POLL RESULTS

What matters more for AI right now?

• Pursuing AGI (general intelligence) (17%)

• Building "AUI" (artificial useful intelligence) (30%)

• Neither—we need to slow down (19%)

• Both equally (20%)

• Something else (write in!) (14%)