First Move  ·  AI Software  · 
Inside the labs — Friday morning, 17 July

AI Labs Graded C+ on Safety by Independent Watchdog; Anthropic Lands Top Talent

Independent assessments highlight AI safety gaps, while labs balance talent acquisition, enterprise expansion, and growing investor scrutiny over mounting debt.

ShareWhatsAppXLinkedIn

The story

The AI software landscape saw a notable shift this week with the release of the Future of Life Institute's 2026 AI Safety Index, which delivered lukewarm grades to major labs. Anthropic received the highest mark, a C+, with OpenAI and Google DeepMind scoring C, and others like Meta, xAI, DeepSeek, and Mistral performing poorly.

The report highlighted a concerning trend of major labs retracting earlier safety commitments, even as their systems become integral to critical sectors like cybersecurity and healthcare. This external scrutiny arrives amidst aggressive talent moves, such as Anthropic reportedly onboarding Andrej Karpathy, a prominent figure from Tesla AI and an OpenAI founding member, and Tom Blomfield, co-founder of Monzo. The concurrent focus on safety accountability and intense talent acquisition underscores the industry's evolving priorities and the tension between rapid development and responsible deployment.

Who moved

Anthropic

What Changed: The company reportedly hired Andrej Karpathy, former Tesla AI director and OpenAI founding member, and Tom Blomfield, co-founder of Monzo, for its AI compute team.

Consequence: This extends Anthropic's aggressive recruiting run in 2026 and solidifies its position in attracting marquee AI and product talent.

Future of Life Institute

What Changed: It released its 2026 AI Safety Index, grading frontier labs on risk management, transparency, and adherence to safety commitments.

Consequence: The highest grade awarded was a C+ to Anthropic, indicating that even the most safety-focused labs are doing a mediocre job by their own stated standards.

Google

What Changed: The company unveiled its expanded Gemini Enterprise platform at Cloud Next.

Consequence: This move reflects a broader industry shift towards coordinated agent systems designed to handle complex business processes with built-in security and oversight.

AI Labs (Singapore)

What Changed: The Singapore-based artificial intelligence trading company launched its AI-powered investment intelligence platform.

Consequence: This marks the company's expansion from AI trading technology into technology equity analysis, bringing institutional-grade market analytics to retail investors.

Market signals

Investors are growing cautious about AI-related debt, which is projected to reach nearly $570 billion globally in 2026.

Implication: This surge in borrowing has not been matched by revenue growth, raising concerns about a potential AI bubble and data center oversupply.

Prediction markets suggest a high likelihood of Anthropic's valuation reaching $1.25 trillion by the end of 2026.

Implication: This indicates strong market confidence in the company's growth trajectory and the broader AI market environment.

Reflection AI, a startup founded by former Google DeepMind researchers, signed a compute agreement worth over $1 billion with Nebius.

Implication: This deal secures access to NVIDIA's latest-generation chips, reflecting the ongoing scramble among AI labs to lock in scarce computing capacity years in advance.

Fintech funding surged 23% in the first half of 2026, with investors concentrating their bets on AI and financial infrastructure.

Implication: This trend highlights strong investor confidence in AI's application within the financial sector, with a focus on large, late-stage deals.

Builders saying

Demis Hassabis, Google DeepMind CEO. He proposed a new US standards body, modeled after financial regulator FINRA, to develop evaluation protocols for frontier models and potentially coordinate AI development slowdowns.

What we'll be watching

Reporting + analyst voices: grounded via Google Search at publish time.