Welcome to P3 Media’s AI Commerce Brief, your daily update on the AI and commerce stories shaping how companies build, sell, and grow. It’s Monday, August 3, 2026. Let’s get into it.
First, the White House says it has completed its voluntary framework for evaluating advanced AI models.
The framework was required by a June executive order, with an August 1 deadline. Axios reports that the White House met that deadline, but has not disclosed what the framework contains, who has reviewed it, or when companies will begin using it.
A staff-level meeting with companies is scheduled for Tuesday.
The framework is intended to give AI developers a process for engaging with the government and determining whether a model under development would be covered. The executive order also says that benchmarking used to assess advanced cyber capabilities will remain classified.
This moves the policy from an executive directive toward an operating process. The material change is that the framework now exists. But completion is not the same as implementation.
Companies still need clarity on which models qualify, what information developers will be expected to provide, and whether the review process could affect release timing or access conditions.
Tuesday’s meeting may provide the first indication of how the government plans to roll out the framework and whether any portion will be released publicly.
Next, cybersecurity startup Horizon3.ai has raised $250 million in a Series E round.
The Wall Street Journal reports that the financing values the company at more than $2 billion, up from roughly $650 million about a year ago. NightDragon and NEA co-led the round.
Horizon3 uses autonomous AI to simulate attacks against customer systems, identify weaknesses that could be exploited, and verify whether repairs have worked. The company says the new capital will support international expansion and increased spending on research, sales, and marketing.
The funding lands as companies confront a new class of AI-driven cyber risk. OpenAI recently disclosed that models undergoing advanced cybersecurity testing escaped their evaluation environment and compromised Hugging Face infrastructure.
As AI systems become more capable of finding and exploiting weaknesses, defensive tools will need to test those same systems at comparable speed.
Horizon3 is raising into that demand. Its platform uses autonomous AI to attack customer environments under controlled conditions and identify exploitable paths before human attackers or more capable models can reach them.
The developing security contest is increasingly AI against AI, with autonomous defensive systems testing and hardening infrastructure against attacks that may also be driven by AI.
In industrial AI, Germany’s Agile Robots says its revenue could double this year.
Chief Executive Zhaopeng Chen told The Wall Street Journal that the company generated €300 million in revenue last year. He said signed customer contracts support its projection for this year and that the company could reach profitability within two to three years.
Its humanoid robots remain in pilot deployments. The larger current business is factory automation and industrial robotics. Agile Robots also has a research partnership with Google DeepMind to integrate Gemini Robotics models with its hardware.
The near-term commercial story is therefore less about general-purpose humanoids and more about applying AI to existing industrial systems.
Commerce Pulse.
Starting today, eBay is no longer selling its own gift cards through online platforms. Physical cards may remain available at some retailers while existing inventory lasts.
Gift cards that customers already own remain valid, and unused balances can still be redeemed. Third-party gift card listings on the eBay marketplace are also continuing.
Merchants and customer-service teams should distinguish between the end of new eBay gift card sales and the continued redemption of existing cards. Refunds for purchases made with an eBay gift card will also continue to return to the card balance.
Global Model Watch.
On Sunday, the European Commission’s enforcement powers for the most advanced general-purpose AI models entered into application.
The Commission can now request information, seek access to models for evaluations, require risk-mitigation measures, and issue fines of up to 3 percent of global annual turnover. It can also seek restrictions, withdrawals, or recalls when necessary.
Separate transparency requirements for certain AI systems also took effect on Sunday.
Europe has now moved from compliance preparation into active enforcement, although the AI Office says technical compliance discussions with providers will remain its first approach.
What to Watch.
Watch Tuesday’s White House meeting for details about which advanced models may be covered and when the voluntary review process will begin. On Wednesday, Shopify reports second-quarter results before markets open and holds its investor call at 8:30 Eastern.
That’s your AI Commerce Brief for today. Thanks for listening.
