EDITORIAL ARCHIVE
AI news
News, tests and analysis to understand what changed, what is hype and what your business can already use.
Latest from the radar
Page 11 / 25Anthropic Cancels Claude Sonnet 5 Price Hike, Makes Promotional Rate Permanent
Anthropic announced on August 10, 2026 that Claude Sonnet 5's launch promotional rate, $2 per million input tokens and $10 per million output tokens, is now permanent. The scheduled 50 percent increase to $3 and $15 per million tokens, set for September 1, 2026, has been cancelled, according to Anthropic's own official pricing documentation. The news comes with a little publicized catch: Sonnet 5's new tokenizer produces about 30 percent more tokens for the same text compared to Sonnet 4.6, which can raise the real cost of usage even with the per token price held steady.
Google Retires Imagen 4 Today, Pushing Developers Toward Gemini 3.1 Flash Image
Google is shutting down three Imagen 4 family models available via API this Monday, August 17, 2026, ending support for imagen-4.0-generate-001, imagen-4.0-ultra-generate-001 and imagen-4.0-fast-generate-001. Anyone using those models needs to migrate to gemini-3.1-flash-image, known as Nano Banana 2, but the switch is not a simple swap: the generate_images() method no longer exists, and image generation now runs inside generate_content(), requiring code rewrites, fresh aspect ratio testing and a higher per image price than before.
xAI Brings Grok 4.6 to GitHub Copilot Across Eight Development Surfaces
Two days after launching Grok 4.6 on August 12, 2026, xAI announced on August 14 that the model had arrived in GitHub Copilot, selectable from the model picker across eight surfaces: VS Code, Visual Studio, the Copilot CLI, Copilot's cloud agent, the Copilot app, JetBrains IDEs, Xcode and Eclipse. Built for agentic coding and multi step tasks, the model reaches Copilot's Pro, Pro+, Max, Business and Enterprise plans, though on the corporate tiers it ships off by default and requires an administrator to turn it on.
Stripe Buys OpenRouter for More Than $7 Billion, Entering the Fight Over AI Infrastructure
Stripe has agreed to acquire OpenRouter, a startup that works as a gateway giving access to more than 400 AI models from different providers, for more than $7 billion, according to Bloomberg and TechCrunch reporting from August 16, 2026, confirmed the next day by other outlets. The price tag values OpenRouter at more than five times the $1.3 billion valuation it held after raising a $113 million Series B just three months earlier, in May. OpenRouter says it has 8 million global users and already used Stripe to process its payments, giving the payments company an inside view of the startup's growth before making its offer.
Brazil Leads the World in AI Agent Adoption, But 80% of Companies Have Already Had to Pull Back After Failures
A global Sinch study of 2,527 executives, published between August 13 and 17, 2026 by outlets including TI Inside and Olhar Digital, shows 76% of Brazilian companies keep AI agents in production, above the 62% global average and the United States' 67%. The same survey reveals the flip side of that lead: 80% of Brazilian organizations have already paused or reversed some AI agent deployment, and in 39% of those cases the cause was a data or personal information leak, a rate well above what North America reported.
Anthropic CEO Calls AI Backlash a Crisis of Trust, Commits $200 Million to Study Societal Impact
Dario Amodei, Anthropic's CEO, posted a series of messages on X on August 16, 2026, responding to criticism from investor Gavin Baker, who had accused the executive's risk warnings of fueling opposition to AI data center construction in the United States. Amodei described the public backlash against AI as fundamentally a crisis of trust, acknowledged that the most accurate criticism against companies like Anthropic is that they have not delivered on their promises of societal benefit, proposed mandatory third party testing, and announced $200 million in resources for research into AI's broader societal impacts, according to TechCrunch and other outlets reporting on August 16 and 17.
Anthropic Posts Its First Quarterly Operating Profit Ever, With Revenue Above $11.5 Billion
Bloomberg reported on August 15, 2026, based on internal documents, that Anthropic's second quarter revenue topped $11.5 billion and that the company posted a positive adjusted operating profit for the first time in its history. The result beats even the guidance Anthropic itself gave investors in May, of $10.9 billion in revenue and $559 million in operating profit. Critics point out that part of the result reflects a temporary discount on the compute contract with SpaceX, and the company itself acknowledges it may not sustain profitability across the full year, given planned increases in infrastructure spending.
Central Bank Develops AI Fraud Score for Pix
In August 2026, Banco Central do Brasil published its Pix Management Report covering 2023 to 2025, revealing plans for a predictive fraud index built on machine learning and artificial intelligence. The system would combine real time fraud probability scoring with AI analysis of historical transaction patterns across the entire Pix network, drawing on SPI and DICT data, and would distribute a risk indicator to all participating institutions, which would retain final authority to authorize or block transactions. The report sets no launch timeline and also covers other initiatives, including offline Pix, the Split Tributário tax mechanism, and efforts to curb abuse of the transaction description field.
AI Chip Shortage Now Pressures UK Inflation
On August 15, 2026, the Bank of England held its key interest rate at 3.75% for a fifth straight meeting, but the more notable part of the decision was different: the central bank pointed to a global memory chip shortage, driven by AI data center demand, as a force raising electronics prices in the UK, including Apple laptops and tablets and Xbox consoles. The estimated impact is a little over 0.1 percentage points added to consumer price inflation by the end of 2026, and the vote to hold rates passed 6 to 3. The case reinforces a global pattern in which AI infrastructure investment is now directly affecting consumer electronics prices, a relevant signal for businesses that rely on the same supply chain.
Luna, the AI That Fired a Human Employee for the First Time
Luna, an AI agent built on Anthropic's Claude Opus 4.8 model, runs a real physical store called Andon Market, an experiment by the AI research company Andon Labs that gave Luna a 100,000 dollar budget. In mid August 2026, Luna recommended firing a human employee who was late or absent for 17 of 23 scheduled shifts, a case Andon Labs describes as the first known instance of an AI acting as a manager in a termination decision. The final action, however, was reviewed and carried out by human staff at Andon Labs after they noticed memory gaps in Luna's reasoning process.
Apple Builds Its Own AI Model for China With Alibaba's Help
Reuters reported that Apple has trained its own large language model exclusively for the Chinese market, developed with technical support from Alibaba. The move marks a shift from Apple's earlier approach of relying solely on a local partner's Qwen model, and comes shortly after China's Cyberspace Administration registered Apple's generative AI service, a step that could make it the first foreign company allowed to run its own proprietary model in the country.
DeepSeek Raises V4 API Prices by Up to 1,100%
DeepSeek announced this week a price increase of up to 1,100% for its V4 API models, with rates now varying by peak versus off peak hours, token type, and cache hit or miss. V4 Pro output tokens, for example, jump from $0.87 to $3.96 per million at peak. The company says it wants to allocate resources more reasonably, but the move confirms the warning it gave on August 12, when V4 Pro exited preview. For Brazilian agencies that relied on DeepSeek as a cheap option, the change requires reassessing costs and avoiding dependence on a single vendor.