The AI price war that started last week is reshaping what frontier models cost, Meta is charging for AI for the first time, and the EU's landmark deepfake labeling law is now 12 days from enforcement.
Frontier AI just got 80% cheaper, and that changes everything
The dust is settling on last week’s simultaneous launches of Grok 4.5, GPT-5.6 (Luna), and Meta Muse Spark 1.1. Output token costs have collapsed from the $25–50 range for legacy flagships to under $6 per million tokens across all three. Developers are now reworking cost models for products they thought were unaffordable to run at scale.
Why it matters: Cheaper frontier AI means products that were only viable for enterprise can now be built by individuals and small teams. If you’ve ever held back on an AI idea because of API costs, the calculation just changed.
Meta starts charging for AI, and that’s the real headline
Meta Muse Spark 1.1 is Meta’s first paid, closed AI model. For years Meta gave its models away through open weights, it was the company’s brand identity. That era is over. Muse Spark 1.1 is priced at $1.25/$4.25 per million tokens, undercutting OpenAI and Anthropic but still a paid product. Analysts say this signals the industry has decided the money is in the inference layer, not the model itself.
Why it matters: Meta going paid is a bigger shift than any model benchmark. The open-weight AI ecosystem that made 2023–2025 so experimentally fertile may be entering a quieter chapter. Watch how the developer community responds.
EU AI Act: 12 days until deepfake labels and chatbot disclosure become law
August 2, 2026 is when the EU AI Act’s Article 50 transparency obligations become enforceable across Europe. Chatbots must clearly state they are not human. AI-generated images, audio, and video must be labeled as artificially made. Emotion-recognition tools must disclose their use to the people being scanned. New compliance guidance was published this week for businesses still finalising their readiness.
Why it matters: This is the first legally binding synthetic-media law at continental scale. Companies operating in Europe must be ready in less than two weeks. Other countries, including India, are watching to see how enforcement actually works.
AI is generating fake job candidates, and HR teams are struggling to spot them
Experian’s 2026 Fraud Forecast names deepfake job candidates as a top emerging threat, with AI-generated applicants capable of passing video interviews in real time using live face-swap and voice synthesis tools. Gartner projects that 1 in 4 job candidate profiles globally will be fake by 2028. Several Illinois-based companies now require AI-disclosure notices in their hiring processes under a law that took effect in January 2026.
Why it matters: The same AI tools being used to help people get jobs are also being used to impersonate job seekers entirely. Hiring managers need new verification steps. Job seekers need to know their authentic track record is now a differentiator.
India raises $346M in one week: AI infrastructure dominates deal flow
Indian startups raised $346.2 million across 20 deals in the week ending July 18, a 332% jump from the same week last year. Yotta Data Services led with a $150M raise at a $4.4 billion valuation, with plans to scale its AI cloud to 85,000 Nvidia Blackwell GPUs by year-end. AI startup Emergent raised $130M from Creaegis, MNI Ventures, and Khosla Ventures. AI cloud and data centre infrastructure are now the dominant funding categories.
Why it matters: India is not just a market for AI, it is now a site of serious AI infrastructure investment. The compute layer being built now will determine who can build what, and where, for the next decade.
India forms a new AI governance body to coordinate national policy
The Government of India has constituted the AI Governance and Economic Group (AIGEG) to build a coordinated, government-wide framework for AI governance, policy coherence, and strategic oversight. The group is intended to prevent fragmented regulation across ministries and bring India’s AI policy in line with its infrastructure ambitions under the IndiaAI Mission, which has already provisioned 38,000+ GPUs for domestic use.
Why it matters: India has the investment and the ambition. AIGEG is the institution meant to make the rules that govern all of it. The decisions this body makes in the next 12 months will shape India’s AI regulatory posture for years.
What happens when AI costs collapse and nobody can afford to ignore it anymore
With frontier AI now available at $4–6 per million output tokens, the last excuse for not experimenting is gone. The next 90 days will reveal which organisations move fast on the new pricing and which ones keep waiting. The gap between those two groups will be measurable by year-end. Watch especially for how Indian startups and mid-size companies respond, because they now have access to tools that only big tech could afford six months ago.
Anthropic drops usage caps + on-device AI keeps growing
Anthropic removed Claude’s 5-hour and weekly usage limits in a direct response to the new low-cost competition, keeping developer loyalty during a week when every major lab dropped prices simultaneously. Separately, on-device AI (models running on phones, laptops, and wearables without a cloud connection) is becoming a real product category, with several startups shipping privacy-first tools that work fully offline. Two trends worth tracking as they compound.