The AI model price war hit a new floor today, fake job applications crossed 1-in-3, and the EU's deepfake labeling deadline is nine days away.
Grok 4.5, GPT-5.6, and Muse Spark 1.1 all landed in 24 hours, and prices are collapsing
xAI's Grok 4.5, OpenAI's GPT-5.6 (codenamed Luna), and Meta's Muse Spark 1.1 all launched within a day of each other, triggering a fresh round of price cuts. Output token costs are now as low as $4–$6 per million tokens, down from $25–$50 for the previous flagship tier just months ago. Frontier AI is getting dramatically cheaper, fast.
Why it matters: This is the price floor dropping in real time. If you use AI tools for work or study, the products you rely on are about to get cheaper and faster. If you build with AI, your API bills are heading down significantly.
OpenAI launched Health in ChatGPT and a new enterprise agent platform
OpenAI officially launched "Health in ChatGPT" on July 23, bringing medical information, symptom guidance, and health planning tools directly into the ChatGPT interface. Separately, OpenAI launched "OpenAI Presence", an enterprise product that lets companies deploy AI agents to answer customer questions, resolve issues, and take approved actions in company systems, escalating to humans when needed.
Why it matters: Health in ChatGPT puts AI in the middle of how people look up medical information, a high-stakes move. OpenAI Presence signals that AI agents are moving from demo to deployed enterprise infrastructure.
1 in 3 job applicant profiles is now AI-generated, and HR teams are struggling
Security firm Pindrop analysed 300 job applicant profiles and found over one-third were entirely fabricated: AI-generated resumes paired with deepfake video interviews. Gartner now projects that 1 in 4 candidate profiles globally will be fake by 2028. Companies report that standard screening is not catching these applications until late in the hiring process.
Why it matters: This is a real and growing problem for anyone involved in hiring, and it's also creating pressure on genuine candidates to prove they are who they say they are. The next screening tool you face in a job interview may be an AI detector.
EU AI Act transparency rules kick in on August 2, nine days from now
The EU AI Act's Article 50 transparency obligations take effect on August 2, 2026. From that date, any AI product available in Europe must inform users when they're interacting with AI, mark AI-generated content in a machine-readable format, and label deepfakes clearly. Companies that miss the deadline face significant fines. In the US, the DEFIANCE Act (which lets victims of non-consensual sexual deepfakes sue creators) is already law.
Why it matters: If you use AI-generated content in your work or business (social posts, videos, marketing materials), and you have any European users or audience, August 2 is a compliance date to take seriously. The world's biggest AI market just set a hard rule.
Etched launches at $5B valuation with $1B in signed chip contracts
AI chip startup Etched launched publicly with a $5 billion valuation and $1 billion in already-signed customer contracts for its specialised AI inference chips. The chips are designed to run large language model inference much more efficiently than general-purpose GPUs, promising lower cost and lower energy use for companies deploying AI at scale. The Federal Reserve also announced a new task force on AI, productivity, and jobs, bringing together economists and AI lab leaders to inform US monetary policy.
Why it matters: The AI hardware race is no longer just Nvidia's to win. Specialised chips for running AI (not training it) are becoming a serious investment category. The Fed's AI task force is the first sign that central bankers are formally treating AI as an economic variable, not just a tech story.
India's MeitY signals a dedicated AI law is coming, and Union Budget backs IndiaAI
India's Ministry of Electronics and Information Technology (MeitY) has signalled that a separate, dedicated AI law is now actively under consideration, with the Secretary stating "the time has come to look at separate legislation for AI." The approach being discussed is risk-based: light rules for low-risk tools like chatbots, heavier obligations for AI used in banking, health, and critical infrastructure. This July's Union Budget allocated ₹551 crore ($66M) to the IndiaAI Mission, funding compute, skilling, startup support, and AI safety pillars.
Why it matters: India is moving from relying on existing IT law to building AI-specific rules. If you're building an AI product for Indian users (or working in a regulated Indian sector) the rules of the road are about to become clearer and stricter at the same time.
Google's flagship model is 8 months late, and they're already building its successor
Google's high-performance Gemini 3.5 Pro (originally planned for June) remains delayed due to underperforming coding benchmarks. During Alphabet's July 22 earnings call, CEO Sundar Pichai confirmed the model is still in testing, while in the same breath revealing that compute is already being allocated to Gemini 4 to compete with frontier rivals. Three lightweight Gemini 3.6 Flash models launched on July 21 to fill the gap. The pattern to notice: at the frontier, even a few months of delay can cost significant market share, and labs are now skipping ahead to the next generation before finishing the current one.
TikTok expands AI literacy to Africa + Google trains educators + Ohio mandates school AI policies
TikTok launched an in-app AI literacy hub across Kenya, Nigeria, and South Africa, teaching users to identify AI-generated content. Google and ISTE+ASCD introduced free, standards-aligned AI literacy courses for K–12 and higher education educators globally. In the US, Ohio has mandated that every public school district adopt a formal AI policy, and July 1, 2026 was the compliance deadline. Schools are now working to catch up.