Anthropic acquisition talks surface | Meta ships Content Seal

Sponsored by

Live Webinar: SSP Automation, Native to Your Billing Platform

Most finance teams still track SSP in spreadsheets — manually updated, disconnected from billing. On July 22, Tabs' product and marketing leads show what changes when SSP runs natively inside your billing platform: built from real billing data, connected to your Product Catalog, and generating audit-ready documentation automatically.

You'll leave with a clear view of how SSP and billing connect, what transaction price allocation looks like under ASC 606 in practice, and what it takes to close faster without reconciling SSP to contracts by hand.

July 22, 2026 · 1:30–2:00 PM EDT · Live + recording

Register for the Webinar

In today's issue

01 · Business

Anthropic held acquisition talks with Physical Intelligence, complicating OpenAI's robotics push

Anthropic held acquisition talks with Physical Intelligence this spring, a $11 billion robotics startup, though the deal didn't close. Both Anthropic and OpenAI are now racing to build robotics capabilities ahead of their planned stock-market debuts later this year.

The details:

Physical Intelligence has raised over $1 billion and makes π0.5, a widely used AI model for controlling robots in academic research.
OpenAI has acquired 17 companies since 2023 and just launched an official robotics lab in San Francisco; Anthropic is testing robot control through its safety research group.
Both companies share investors in Physical Intelligence, including Khosla Ventures, Thrive Capital, and Founders Fund, creating overlap in their robotics strategies.

Why it matters: Understanding the physical world may be essential for the next generation of AI systems, and buying an existing robotics team lets a company skip years of building hardware and collecting real-world data. The competition between Anthropic and OpenAI to own robotics talent will likely intensify as both prepare to go public.

Read the full story →

02 · Security

Meta launches Content Seal AI watermarking, three years after promising labels

Meta launched Content Seal in July, an invisible watermark that identifies images made by its new Muse model—three years after promising to label AI-generated content. The watermark only works on Muse images, leaving millions of AI photos Meta generated since 2023 undetectable by its own tool.

The details:

Detection only runs through a rate-limited web tool, not inside Meta AI where users actually see AI images, and the company hasn't said when or if it will add detection to its chatbot.
Content Seal copies Google's SynthID (which OpenAI already uses), but Meta built its own standard despite sitting on the steering committee for C2PA, an industry-wide standard.
When The Verge tested a Muse image with Google's Gemini and the official C2PA detector, neither recognized it as AI-generated, raising questions about whether Meta's watermark works across platforms.

Why it matters: Meta's three-year delay and half-built rollout suggest the company is checking a box rather than solving the problem. Until detection lives inside Meta AI itself and covers everything the company has generated since 2023, the watermark won't actually stop deceptive AI content from spreading on Facebook and Instagram.

Read the full story →

03 · Business

Synthesia launches Roleplay Sessions to score employees on AI conversations

Synthesia launched Roleplay Sessions on Wednesday, letting employees practice conversations with AI avatars that grade them against a rubric. The British startup is betting enterprises will pay more for proof that training actually changes behavior, not just cheaper video clips.

The details:

Early customers include one of Europe's top three companies by market value, a top-five Fortune 100 firm, and one of the world's largest recruitment companies.
The product uses OpenAI's reasoning engine for grading, while Synthesia keeps its avatar and voice technology proprietary to stay competitive.
Synthesia plans to expand from enterprise-only to small businesses and schools within months as the cost to run AI conversations keeps dropping.

Why it matters: This positions Synthesia as a performance-management platform rather than just an AI video vendor. The shift makes sense: most corporate training stops at watching and learning, but salespeople and managers get better through repeated practice with feedback. If Roleplay's analytics can prove measurable improvement in skills like sales or leadership, enterprises will pay premium prices for it.

Read the full story →

AI Box

Every AI model. One chat.

The latest models from ChatGPT, Claude, Gemini, Sora, ElevenLabs — 80+ models in a single chat. Compare answers side by side. Pick the best one every time.

● ChatGPT, Claude, Gemini, Grok, DeepSeek — in one chat

● Generate images & video with Sora, Veo, Ideogram

● Compare any two models side by side

● From $8.99/mo · 80+ models, all included

Try AI Box →

aibox.ai

Trusted by 3,000+ teams

04 · Business

Samsung in talks to back Mistral at €20 billion valuation

Samsung is in talks to invest in French AI startup Mistral at a €20 billion valuation, more than triple its value from mid-2024. The deal would cement Mistral as Europe's only frontier AI lab competing with OpenAI and Anthropic, while giving Samsung a direct stake in a model maker it already uses to power Galaxy phones.

The details:

Mistral raised at €5.8 billion in June 2024 with General Catalyst leading; a €20 billion mark would be a 3.4x jump in under 18 months, matching US frontier lab growth rates.
Samsung already integrates Mistral models into Galaxy device features in select markets to reduce reliance on Google's Gemini and offer region-specific choices.
Mistral was founded in 2023 by former Meta and DeepMind researchers; it focuses on open-weight models and API services for European enterprises navigating the EU AI Act and data sovereignty concerns.

Why it matters: This positions Mistral to compete harder in Asia and reinforces its appeal to European governments and enterprises as a non-US alternative. Samsung joins Nvidia and SoftBank in using hardware and cash to secure preferential access to frontier AI models, a trend that's reshaping how models get distributed globally.

Read the full story →

05 · Business

NextEra, Duke, and 200 utilities sign Trump's AI electricity pledge

Nearly 200 utility and data center companies—including NextEra Energy and Duke Energy—have signed Trump's pledge to shield households from AI electricity costs, now covering 80% of US power delivery. The voluntary commitment aims to make AI providers, not regular ratepayers, pay for new power plants and grid upgrades.

The details:

Original signers from March 2026 included Google, Meta, Microsoft, Oracle, OpenAI, and Amazon; the expanded list adds major utilities and data center operators like Equinix and Digital Realty.
PJM, the largest US grid operator, is already passing $6.3 billion in new costs to consumers across 13 states due to data center demand, independent of the pledge.
The pledge carries no penalties for non-compliance and cannot override state utility commissions, which actually set the electricity rates households pay each month.

Why it matters: The pledge is a political alignment play—tech companies and utilities get goodwill at a moment when data center approvals face local pushback, but it doesn't change how power purchase agreements work or stop costs already flowing into rate cases. Without federal rate-setting authority or enforcement teeth, the real test is whether it changes behavior at the state level where regulators actually decide prices.

Read the full story →