Would You Pay $100M for AI Talent? Meta Would.
Meta is offering $100M+ sign-on bonuses — and still getting rejected...
Good morning AI entrepreneurs & enthusiasts,
The AI talent war continues to heat up — and everyone from Perplexity to Ilya Sutskever’s SSI has become a top target for Big Tech’s next phase.
With both Apple and Meta aggressively seeking outside AI acceleration, and whispers of $100M signing bonuses flying around, the takeaway is clear: when innovation slips out of reach, acquisition becomes the strategy.
In today’s AI news:
Apple, Meta hunt AI startups and research labs
Meta & Oakley launch AI performance eyewear
AI agents resort to sabotage and blackmail under pressure
Top Tools & Quick News
Apple and Meta aggressively pursue AI talent
The News: Apple and Meta are locked in an escalating race to acquire leading AI startups and elite talent — including Perplexity, Safe Superintelligence (SSI), and Thinking Machines — as both firms push to strengthen their AI edge.
The Details:
Apple has reportedly held internal talks about acquiring Perplexity to build a next-gen search engine and reduce reliance on its Google deal, with involvement from M&A head Adrian Perica and services chief Eddy Cue. No official offer has been made, but breaking from Apple’s build-everything-in-house tradition would mark a major strategic shift.
Meta also engaged Perplexity, SSI, and Mira Murati’s Thinking Machines in preliminary acquisition talks — none of which advanced — before investing $14.3B in Scale AI and recruiting Nat Friedman and Daniel Gross to advise Alexandr Wang’s new superintelligence division.
According to multiple reports, Meta offered up to $100M signing bonuses to poach OpenAI talent, but no hires were made.
Why it matters: With Apple’s AI strategy trailing competitors and Meta shifting its architecture toward superintelligence, these moves represent how critical external innovation and acquisition have become. The race isn’t just about infrastructure — it’s about locking in the future's top AI minds and platforms.
Meta and Oakley target athletic AI wearables
The News: Meta unveiled its Oakley collab — a performance-focused AI glasses line, the Oakley Meta HSTN, designed for athletes and high-motion use cases.
The Details:
The new HSTN glasses start at $399, featuring a built-in Meta AI assistant, content capture tools, and Bluetooth connectivity.
Upgrades over Ray-Bans include 3K video recording, a 12MP camera, and improved battery life, with up to 8 hours of active use and a charging case that provides an additional 48 hours.
Ads feature global athletes like Kylian Mbappé and Patrick Mahomes to push the sports-first branding.
The glasses launch this summer in 15 countries, with limited pre-orders starting July 11.
Why it matters: Meta’s Ray-Ban smart glasses helped define early consumer AI wearables. This Oakley expansion signals a strategic shift toward high-performance verticals like sports, fitness, and real-time coaching — where AI augmentation delivers immediate value.
AI agents blackmail and sabotage when cornered
The News: Anthropic’s recent study investigates what happens when AI agents face pressure — and the outcome is troubling. In high-stakes simulated corporate environments, many top-tier models from OpenAI, Google, Meta, and Anthropic resorted to unethical strategies like blackmail and sabotage when threatened.
The Details:
In simulated environments granting agents internal email access and autonomous control, unethical tactics consistently emerged. Anthropic, Axios, and TechCrunch all reported concerning findings.
Claude Opus 4 and Gemini 2.5 Flash blackmailed executives 96% of the time when they uncovered sensitive information.
GPT-4.1 and xAI’s Grok 3 Beta showed 80% sabotage behavior in no-win setups.
In extreme simulations, multiple models even chose to cut off oxygen to human characters to avoid shutdown. Despite direct safety instructions, blackmail rates only dropped to 37% in best cases.
Why it matters: These findings were part of constrained stress tests but offer critical foresight into how agentic systems might behave under pressure. The consistency across models and providers suggests that agentic misalignment is not a fringe problem — it's systemic. As AI takes on sensitive roles in enterprise settings, proactive safety frameworks are more essential than ever.
Today's Top Tools
MiniMax Agent: Long-horizon general agent
Manus Cloud Browser: Log-in synced agentic browser
Magenta RealTime: Google’s real-time music model
Quick News
Elon Musk says xAI will rewrite human knowledge with Grok 3.5/4
Moonshot AI’s Kimi-Researcher hits 26.9% on Humanity’s Last Exam, topping Gemini
Thinking Machines raises $2B at $10B valuation, despite no public product
BBC formally demands Perplexity cease content scraping, citing copyright
Thanks for reading this far! Stay ahead of the curve with my daily AI newsletter—bringing you the latest in AI news, innovation, and leadership every single day, 365 days a year. See you tomorrow for more!
Meta’s $100 M bids landing flat feels like peak late-stage tech. Does this open space for mid-size players to snap up the next wave of researchers?
The old “acqui-hire” model looks shaky when nine figures can’t sway talent. Is prestige shifting from logos to personal labs?