The first week of August 2026 brought four developments that reshape how AI gets regulated, priced, and packaged. Europe's transparency rules became enforceable law. OpenAI cut prices sharply in response to competitive pressure. Microsoft moved to unify its AI products into a single app. And Anaconda made a notable acquisition in the AI security space.
This Week in AI — Episode 16 covers four stories from August 1–7, 2026: the EU AI Act's transparency obligations took effect on August 2, requiring AI systems to disclose when users are talking to a chatbot and to label AI-generated content. OpenAI cut GPT-5.6 Luna's input token pricing by 80% on July 30, intensifying competition with lower-cost open-weight models. Microsoft confirmed at its Q4 earnings call that Copilot chat, Code, Cowork, and Autopilots are merging into one unified app this quarter. Separately, data science platform Anaconda acquired AI security startup Enkrypt AI to add pre-deployment testing and runtime protections for enterprise AI systems.
Each of these stories points to the same underlying shift: AI is moving from an experimental technology to regulated, competitively priced, consolidated infrastructure. Below is a breakdown of each story and what it signals for where the industry is headed next.
1 What Do the EU AI Act's New Transparency Rules Require?
As of August 2, 2026, AI providers operating in the EU must meet the Act's first enforceable transparency requirements — the earliest concrete compliance deadline of the law's rollout.
Three obligations are now active: chatbots and AI assistants must clearly disclose that users are interacting with AI rather than a human; synthetic or manipulated media (deepfakes) must be labelled as AI-generated; and AI-generated content more broadly must carry machine-readable markers so platforms and downstream tools can identify it. Non-compliance carries fines of up to €15 million or 3% of a company's global annual turnover, whichever is higher.
Notably, this is a narrower rollout than the Act's full scope. The harder-edged rules — governing "high-risk" AI systems used in hiring, credit decisions, and critical infrastructure — were separately postponed to December 2027, giving companies more runway to build out compliance systems for those higher-stakes categories.
This is the first time a continent-wide AI law has moved from policy to enforceable obligation. Even companies outside the EU that serve European users will need to build disclosure and labelling into their products, making this a de facto global baseline rather than a regional one.
Live now: chatbot disclosure, deepfake labelling, machine-readable AI content markers. Postponed to December 2027: obligations for high-risk systems in hiring, credit, and critical infrastructure.
2 Why Did OpenAI Cut GPT-5.6 Luna's Price by 80%?
OpenAI cut Luna's input token pricing from $1 to $0.20 per million tokens on July 30 — just three weeks after the model's launch — in what analysts are describing as the opening move of an AI price war.
The cut appears to be a direct response to competitive pressure: Chinese open-weight models have been capturing an increasing share of US enterprise token usage, reportedly around 46% on aggregator platforms like OpenRouter. Terra, another OpenAI model tier, also dropped in price by roughly 20%, while Sol's pricing held steady — suggesting a targeted, tier-specific response rather than a blanket repricing.
What to watch: If competitors respond with their own cuts, input-token pricing across the industry could compress quickly. For businesses building on these APIs, this is a good moment to revisit cost assumptions in any AI-dependent product roadmap.
3 What Is Microsoft's Copilot "Super App"?
Microsoft confirmed during its July 29 Q4 earnings call — which showed $90 billion in revenue, up 18% year-over-year, with the stock rising 15% — that it is merging Copilot chat, Copilot Code, Copilot Cowork, and its Autopilots into a single unified application launching this quarter.
CEO Satya Nadella framed the move as consolidating what have been separate Copilot experiences into one entry point for users, regardless of whether they're chatting, coding, or delegating multi-step work. The announcement puts Microsoft in direct competition with OpenAI's ChatGPT Work and Anthropic's Claude Cowork, both of which pursue a similar "one app for everything" strategy for professional AI use.
Copilot chat, Code, Cowork, and Autopilots are merging into one unified app launching this quarter.
— Satya Nadella, Microsoft Q4 FY2026 earnings call
The race toward a single, unified AI "super app" is now a three-way contest between Microsoft, OpenAI, and Anthropic — each betting that consolidating fragmented tools into one interface is what enterprise and professional users actually want.
4 Why Did Anaconda Acquire AI Security Startup Enkrypt AI?
Data science platform Anaconda announced on August 4 that it has acquired Enkrypt AI, a startup focused on pre-deployment testing and runtime protection for AI systems and AI agents.
The deal folds Enkrypt AI's testing and monitoring capabilities directly into Anaconda's existing data science and AI development platform, giving Anaconda's enterprise customers built-in tooling to evaluate AI systems before deployment and monitor them once they're live. Anaconda cited the fast-growing footprint of AI agents in enterprise workflows — and the corresponding need for organizations to test and monitor those systems before and after rollout — as the rationale for the acquisition.
Anaconda's move mirrors a broader pattern this year: established data and developer platforms acquiring specialized AI safety and testing startups rather than building those capabilities from scratch, folding governance and testing directly into the tools developers already use.
What This Means for You
Taken together, this week's stories mark AI's shift from an experimental technology to regulated, competitively priced, and increasingly consolidated infrastructure. Regulation is now enforceable, not theoretical. Pricing is being driven down by real competition, not just marketing. And the biggest platforms are converging on the same "one app" strategy. Staying curious about these shifts — and understanding how they connect — is exactly the kind of exploration we're here for. Stay Curious. Explore What's Next. Grow Together.
Starting August 2, 2026, AI providers serving EU users must disclose when someone is interacting with a chatbot rather than a human, label AI-generated deepfakes, and add machine-readable markers to AI-generated content. Fines for non-compliance can reach €15 million or 3% of global turnover.
OpenAI cut Luna's input token price by 80% on July 30, 2026, just three weeks after launch, largely in response to Chinese open-weight models capturing a growing share of US enterprise AI usage. It's widely seen as the start of a broader industry price war.
Announced at Microsoft's Q4 FY2026 earnings call, the Copilot super app merges Copilot chat, Copilot Code, Copilot Cowork, and Microsoft's Autopilots into a single unified application launching later in 2026, positioning it against ChatGPT Work and Claude Cowork.
Anaconda acquired Enkrypt AI on August 4, 2026, to add pre-deployment testing and runtime monitoring for AI systems and AI agents directly into its data science platform, responding to growing enterprise demand for built-in AI governance tools.