IIT Hyderabad and IIT Roorkee Launch AI Training Programs in 2026

Key Takeaways

• IIT Hyderabad launched a three-month AI engineering programme starting December 5, 2026, for working professionals.
• IIT Roorkee's EICT Academy is offering five AI courses at Rs 8,000 each, with batches starting December 14, 2026 and January 4, 2027.
• Investors warn the AI investment boom shows bubble signs, with the Buffett Indicator at roughly 240 percent, above dot-com bubble levels.
• Circular financing patterns seen in the dot-com crash are reappearing in today's AI sector, involving Anthropic, Nvidia and Microsoft.
• CoreWeave reported Q3 2026 sales of $3.66 billion but spent $9.4 billion on equipment and buildings, with a $132.5 billion backlog.
• Disney and Adobe are building custom generative AI models through Firefly Foundry, using Disney's franchises for proprietary model training.
• Meta launched the Muse Charm, a keychain-size AI device that encourages users to connect personal accounts including email and financial data.
• A study on agent memory evaluation found Qwen3-32B matched 87 of 85 hidden references and DeepSeek-V4-Flash matched 85 on 80 confirmation histories.
• Insurance markets are responding to AI risk, with Anthropic's record IPO near a US$1.4 trillion valuation prompting scrutiny of insurability.
• Nvidia stock faced pressure from the CEO's affair saga but retains strong competitive advantages amid election-driven market volatility.

IIT Hyderabad launches 3 month AI engineering programme

IIT Hyderabad has opened registrations for a three month Forward Deployed Engineering Practitioner Programme. The course starts December 5, 2026 and ends March 14, 2027 with 52 sessions. It is designed for working professionals with technical backgrounds and focuses on building practical AI systems for real business use.

IIT Roorkee launches 6 week winter AI training internship

IIT Roorkee's EICT Academy is offering a six week winter training and internship in AI. The programme costs Rs 8,000 and will run online in two batches starting December 14, 2026 and January 4, 2027. It covers five courses including Generative AI, Machine Learning, Data Science, Full Stack Web Development and IoT. Applications close on November 30, 2026.

IIT Roorkee opens registration for five AI courses

IIT Roorkee has started registration for five short term AI courses through its EICT Academy. The courses include Generative AI and Prompt Engineering, Machine Learning and Agentic AI, Data Science using Python, AI-Powered Full Stack Web Development and IoT and Edge AI. The fee is Rs 8,000 and classes will be held online. The first batch begins December 14 and the second starts January 4.

Experts warn AI investment boom shows bubble signs

Veteran investors at the New Orleans Investment Conference warned that the AI investment boom shows many signs of a market bubble. They pointed to high valuations, rising leverage and speculative capital flows as warning signs. The Buffett Indicator has reached about 240 percent, well above the dot-com bubble level. Experts say recognizing a bubble is easier than knowing when it will burst.

Dot-Com era lessons offer warnings for the AI bubble

A commentary draws parallels between the current AI investment boom and the dot-com bubble of the late 1990s. It highlights how infrastructure companies relied on heavy debt and circular financing during the dot-com crash, with many going bankrupt. The article notes similar circular financing patterns in today's AI sector involving companies like Anthropic, Nvidia and Microsoft. It warns that trillions in AI infrastructure spending may be hard to justify without massive revenue growth.

3 insurance stocks exposed to AI risk

A new AI shock is rippling through insurance markets as Anthropic confirms its systems are accelerating their own R&D and lines up a record IPO near a US$1.4 trillion valuation. Sudden talk of uninsurable AI accidents meets soaring demand for the chips and cloud providers funding that growth. For investors, the question is which insurers might quietly benefit. This piece walks through 3 stocks exposed to this news and explains why their positioning matters now. The three stocks below are just a sample of the insurers and reinsurers exposed to large-scale technology risk. The full screen surfaced 8 more companies with equally compelling narratives that are not covered here. To identify your own highest-conviction ideas, analyze IRB-Brasil Resseguros, a Brazilian reinsurer in the technology and cyber exposed segment of this screener, writing a broad mix of casualty, property, life and specialty protection and generating about R$4.9b in reinsurance operations revenue, with a market value near R$4.7b.

Companies may be held liable when AI harms clients

Artificial intelligence has moved quickly from a novelty on the job to a common workplace tool. Gallup found that 52% of U.S. employees and 68% of professional services workers were using AI in some capacity. However, as AI becomes more prevalent, concerns about its impact on workers and customers are growing. The use of AI in the workplace has been shown to have both positive and negative effects. On one hand, AI can help automate tasks, freeing up human workers to focus on more complex and creative tasks. On the other hand, AI can also lead to job displacement and increased stress for workers. In addition, AI can also have a negative impact on customers. For example, AI-powered chatbots can sometimes provide inaccurate or incomplete information, leading to frustration and mistrust. Furthermore, AI can also perpetuate biases and stereotypes, which can have serious consequences. As AI continues to evolve, it is essential that companies take steps to mitigate its negative effects and ensure that their AI systems are fair, transparent, and accountable.

Controlled study of agent memory evaluation

An answer can match the reference for a complete agent history even when the reader's visible history does not determine it. The study uses executable device logs to separate reference agreement, certified evidence sufficiency, and context-appropriate behavior. Insufficiency requires two legal histories yielding byte-identical visible contexts but different answers; public chronology and query cutoffs certify sufficient prefixes. On 80 confirmation histories, Qwen3-32B matches 87 hidden references among 203 insufficient raw-prefix instances, with only seven appropriate abstentions. A fixed-input DeepSeek-V4-Flash replication matches 85 and abstains on 21 of the same 203 instances. Each reader makes 1,600 requests. Compact CSV's estimated appropriateness effects remain inconclusive: +3.16 percentage points, clustered 95% interval [0.00,6.57], for Qwen and +2.46, [-1.84,6.94], for DeepSeek. A separate held-out judge study achieves 99.38% correctness but retains support errors and truncations; a deterministic baseline solves all constructed cases.

Cute faces like Muse Charm are a Trojan horse for exploitative AI tech

Mark Zuckerberg of Meta introduces the Muse Charm device in September. This is a keychain-size gadget that looks like something you'd find in a child's bedroom, a distant cousin of Furby or a Pixar character. It's soft and approachable. Nothing about its appearance suggests the power of the technology behind it, or the consequences of giving it access to our private lives. Meta has rolled out its new AI assistant, the Muse Charm. The marketing feels deliberately disarming. Muse Charm encourages users to connect email, messages, photographs and financial accounts. Through that process, we're creating detailed records of our private lives, with few guarantees about how that information will be protected. In my work representing survivors of sexual exploitation, stalking and technology-facilitated abuse, I've seen how devastating it can be when intimate information falls into the wrong hands. My firm has represented survivors whose intimate images were distributed online without consent, sometimes alongside identifying information that made it easier for strangers to find and harass them. Reclaiming control often takes years. The same pattern is emerging with increasingly powerful technologies that make it easier to invade someone's privacy and turn their personal information against them.

Disney and Adobe are building custom generative AI models

Disney and Adobe are building custom generative AI models for parks and ESPN. Adobe will give Disney early access to Firefly, Premiere and Photoshop under a multi-year alliance announced Monday. Through Firefly Foundry, Disney will collaborate with Adobe on proprietary generative AI systems built around its own franchises and properties. The models learn from character attributes, environments and visual styles, which lets them produce output that is safe for commercial use and ready for production. Walt Disney Imagineering Research & Development has used Firefly Foundry in theme park development, Adobe said. Its tools include a sketch-to-image model that takes artists' hand-drawn ideas and renders them as 2D concept art, plus a custom image model that produces assets faithful to the franchises. The work will begin by shortening project schedules in Disney's Experiences division and supporting early concept work at StudioLAB, Walt Disney Studios' internal innovation arm, according to.

Nvidia and other strong stocks may offer opportunities after election selloff

Recent election-driven market volatility has pushed fundamentally strong stocks to discounted prices. Nvidia faced stock pressure due to the CEO's affair saga but retains strong competitive advantages. The S&P 500 trades at a forward P/E of 17.7x, its lowest in three years. Other companies like Lockheed Martin and Amazon also show strong growth and profitability metrics.

CoreWeave beats sales forecasts while spending billions on data centers

CoreWeave reported third quarter 2026 sales of $3.66 billion, up 168% from a year earlier and above its forecast. The AI cloud company spent $9.4 billion on equipment and buildings, which is 2.6 times its sales. Its backlog of signed work reached approximately $132.5 billion. CoreWeave's spending outpaces competitors like Microsoft and Meta relative to sales.

New research introduces algebraic tree decision diagrams with proven canonicity

Researchers introduced algebraic tree decision diagrams (ATDDs) as a new way to represent Boolean functions. ATDDs use the rank of a cut matrix instead of counting subfunctions and are exponentially more succinct than traditional tree decision diagrams. The study proves ATDDs are the first strict canonical form for rank-based representations over arbitrary vtrees. Core results were formalized and verified in Lean, with a prototype implementation also provided.

Sources

• IIT Hyderabad launches 3 month AI Engineering programme for working professionals: Check course details, fees and eligibility
• IIT Roorkee Offers 6-Week Winter Training And Internship In AI: Apply By November 30
• IIT Roorkee offers winter training and internship in these 5 AI courses
• AI Investment Frenzy: Bubble Warnings from Experts
• Dot-Com lessons for the AI bubble
• 3 Insurance Stocks for Rising AI Risk in Cyber and Specialty Cover
• Companies may be held liable when AI harms their clients and customers
• Correct Answers from Insufficient Histories: A Controlled Study of Agent Memory Evaluation
• Contributor: Cute faces like Muse Charm are a Trojan horse for exploitative AI tech
• Disney and Adobe are building custom generative AI models for parks and ESPN
• Nvidia's AI Moat Faces A New Test, And Other Top Stocks With Durable Earnings
• AI Computer Renter CoreWeave Beat Its Own Sales Forecast and Spent $9.4 Billion on Data Centers
• Factor Rank: A Canonical Algebraic Generalization of Tree Decision Diagrams