Business & Funding 26 October 2026
Anthropic files to list its shares on the stock market
Anthropic, the company behind the Claude AI models, filed for an initial public offering. It plans to trade under the ticker 'ANT'. The filing shows revenue of $16.3 billion for the first half of 2026, up from $1.2 billion a year earlier. It also shows a net loss of $80.7 billion, which the company says is mostly an accounting charge linked to convertible debt. Anthropic has not yet said what price it will ask for its shares or how much money it wants to raise.
Why it matters: Many Indian companies and developers use Claude. As a listed company, Anthropic would have to publish detailed figures every quarter, so customers can see how stable a key AI supplier is.
Models & Products 26 October 2026
OpenAI releases GPT-6.1 Terra at half the price of the model it replaces
OpenAI released GPT-6.1 Terra, a fast mid-range model built for operating a computer and understanding images. It replaces GPT-5.6 Terra. On OpenAI's API it costs $1 per million input tokens and $5 per million output tokens, which OpenAI says is 50% lower than before. Paid ChatGPT users can use it in ChatGPT Work and Codex. OpenAI rates its cybersecurity ability as 'Critical' under its own safety framework.
Why it matters: Indian startups and businesses that build on OpenAI can cut costs for routine, high-volume tasks. Before switching, they should test it on their own work, because the performance figures come from OpenAI.
Models & Products 28 October 2026
Google announces Gemini 4 Carbon, first for cyber defenders only
Google announced Gemini 4 Carbon and called it its strongest model for cybersecurity and long tasks. For now only a small group of cyber defenders can use it. Google will widen access to trusted partners through a scheme called the Fairwind Program, and has given no date for wider release. Google says the model does well on tests of finding, exploiting and fixing software flaws. It also says Carbon is more than twice as likely as Gemini 4 Argon to write secure code.
Why it matters: Google is limiting who can use this model because it is so good at finding software weaknesses. Indian businesses should expect such tools to reach attackers too, and should keep their software updated.
Safety & Security 27 October 2026
Anthropic shows an AI model building working attacks on major software
Anthropic's safety team reported that an unreleased internal AI model found serious flaws in Chrome, Safari, Android, Windows and Linux. The model linked flaws together into five working attacks, with little human help. In Chrome alone, 62 flaws received official vulnerability numbers, all rated critical or high. Anthropic says the flaws were reported privately to the software makers and have been fixed.
Why it matters: The time between a flaw being found and an attack being built is getting shorter. Indian users and IT teams should install browser, phone and operating system updates quickly.
Safety & Security 28 October 2026
Flaw in shared ChatGPT links could expose private chats; OpenAI has fixed it
Researchers at the security firm Parameter published details of a flaw in ChatGPT's shared-chat links. A harmful link could run an attacker's code in the session of a signed-in user who opened it, and read that user's private chats. OpenAI fixed the problem on its own servers by October 26 and paid the researchers $5,000. The researchers reported no real-world attacks.
Why it matters: Users do not need to do anything. Many Indian professionals paste business or personal data into chatbots, and this shows why it is safer to be careful about what goes into them.
Business & Funding 28 October 2026
Microsoft lowers its AI spending forecast; Alphabet shares fall
In its quarterly results, Microsoft cut its full-year spending forecast from $175 billion to $170 billion. Its finance chief said delays by AI chip suppliers will push some capacity into the first half of 2027. Alphabet kept its 2026 spending plan at $195 billion to $205 billion. It reported negative free cash flow of more than $9 billion, and its shares fell about 10% in after-hours trading.
Why it matters: Chip delays at the largest cloud companies can affect how much AI computing power is available and what it costs, including for Indian firms that rent cloud capacity. Investors are now looking harder at whether AI spending pays off.
Business & Funding 30 October 2026
OpenAI reported to cut its funding target to $20 billion
Last week OpenAI was reported to be close to raising up to $30 billion. The Financial Times now reports that it has cut the target to about $20 billion, at a valuation of $1.25 trillion instead of $1.4 trillion. The report says some investors reduced their commitments after AI shares fell. OpenAI declined to comment.
Why it matters: Even the largest AI companies now find it harder to raise money. Indian AI startups may face investors who ask more questions about costs and revenue.
India 31 October 2026
MeitY proposes a consent-based way to share data for training AI
The IT ministry proposed a framework called DEPA for AI Training, developed with the think tank iSPIRT. It builds on the consent-based system already used by Account Aggregators. Organisations such as banks, hospitals and government agencies could let others train AI models on their data inside secure 'clean rooms' without handing over the raw data. Models trained this way would have to pass checks for bias, safety and security before being used.
Why it matters: If it is adopted, Indian startups and researchers could get large Indian datasets to build AI in Indian languages and for Indian needs. Citizens' personal data would stay protected under the DPDP Act.
India 30 October 2026
DSCI releases a guide on who is responsible when AI agents cause harm
The Data Security Council of India, an industry body set up by Nasscom, released a guide on liability for 'agentic' AI, meaning AI that takes actions on its own. It does not propose new rules. It explains how existing Indian law applies when several companies supply parts of an AI system. DSCI calls it a live discussion paper that will be updated with feedback from industry.
Why it matters: Indian firms that use AI agents for payments, customer service or operations can use the guide to set out in contracts who is responsible if the AI makes a costly mistake.
Safety & Security 30 October 2026
RBI reported to step up cyber checks on banks over AI-driven threats
A news report says the Reserve Bank of India has increased its oversight of banks as cyberattacks on the financial system rise. According to the report, the RBI wants proof that banks follow existing cyber rules and are ready for AI-related risks. It has also asked them to check weak points in the APIs that connect them to NBFCs and fintech firms. The RBI did not immediately comment on the report.
Why it matters: Smaller lenders, NBFCs and fintech partners may face tighter security checks. Customers should stay alert to fraud, because criminals are using AI to make attacks more convincing.
India 28 October 2026
Gnani AI launches a speech-to-text model for Indian phone calls
Bengaluru-based Gnani AI launched Prisma v3.0, a model that turns speech into text. It works in nine Indian languages plus English: Hindi, Bengali, Marathi, Gujarati, Punjabi, Kannada, Telugu, Tamil and Malayalam. It is trained on noisy phone-call audio and on speech that mixes languages. Developers can use it through Gnani's API.
Why it matters: Call centres, banks and service businesses in India could transcribe customer calls in local languages more easily. The company has not published an independent accuracy test, so buyers should run their own trials.
Business & Funding 27 October 2026
Nvidia invests in Bengaluru startup Simplismart
Simplismart, a Bengaluru startup, raised $14 million in a Series B round led by Dallas Venture Capital. Nvidia, Accel, Shastra VC and Micromax also invested. The company's software helps businesses run AI models faster and with fewer GPUs. Its customers include InVideo and the Indian cloud provider Yotta, and the deal values the company at about $100 million.
Why it matters: GPUs are expensive and hard to get in India. Software that gets more work out of the same hardware can lower the cost of AI services for Indian businesses.
Compiled from public reports and official announcements. Follow the source links for full details. Next update: Saturday, 7 November 2026, 2:00 PM IST.