anthropic, openai and google Updates

Recent developments in AI reveal a complex landscape of successes, failures, and future ambitions. An MIT report indicates that a staggering 95% of AI pilot projects fail to generate profit, often due to misuse and poor planning rather than the technology itself. This is leading companies to consider buying AI solutions instead of building their own, and to focus on back-end applications. Despite these challenges, Anthropic's Claude is gaining traction in the enterprise sector, surpassing OpenAI's ChatGPT with a 32% market share compared to ChatGPT's 25%, while Google holds 20%. Coding is identified as a key area where businesses are seeing returns on their AI investments. Meanwhile, OpenAI is planning trillion-dollar data centers to support its future AI infrastructure needs, aiming for scalability and cost-effectiveness. Meta is also heavily investing in AI, allocating $72 billion to develop 'personal superintelligence' and seeing revenue boosts of $46.56B in Q2 2025 from AI-driven ad systems. However, concerns are rising about the misuse of AI, with experts warning against using AI chatbots like ChatGPT for health advice, citing potential for incorrect information and addiction. Scams involving AI are also on the rise, as demonstrated by a man in Hyderabad losing Rs 2.36 crore through a WhatsApp-based AI trading scam. On the hardware front, Roborock is working to make AI-powered robot vacuums more affordable, while the University of Nebraska-Lincoln is opening a new AI lab equipped with NVIDIA H100 GPUs. Reolink is set to showcase its Next-Gen AI Security System and a new 4K camera at IFA 2025. Finally, new legal standards are being proposed to regulate the use of AI-generated evidence in court, ensuring reliability and accountability.

Key Takeaways

• An MIT report reveals that 95% of AI pilot projects fail to generate profit due to misuse and poor planning.
• Anthropic's Claude AI model leads enterprise use with a 32% market share, surpassing OpenAI's ChatGPT (25%) and Google (20%).
• OpenAI plans to build trillion-dollar AI data centers to secure computing power for future AI development.
• Meta is investing $72B in AI to create 'personal superintelligence,' with AI-driven ad systems boosting revenue to $46.56B in Q2 2025.
• Experts warn against using AI chatbots like ChatGPT for health advice due to potential for incorrect information and addiction.
• A man in Hyderabad lost Rs 2.36 crore in an online scam through a WhatsApp group promising AI-based trading algorithms.
• Roborock plans to make AI-powered robot vacuums more affordable, aiming for human-like robots to be common by 2030.
• The University of Nebraska-Lincoln is opening a new AI lab with NVIDIA H100 GPUs, supported by Scott Data.
• Reolink will present its Next-Gen AI Security System and a new 4K camera at IFA 2025.
• A proposed Rule 707 would set standards for AI evidence in court, requiring it to meet the same standards as expert testimony.

MIT report reveals AI pilot projects failing due to misuse

An MIT report found that 95% of AI pilot projects don't make money or increase profits. Investors are worried, but the report says the problem isn't the AI itself. Instead, companies don't know how to use AI tools correctly or design good workflows. The report suggests companies should buy AI solutions instead of building their own. Also, AI could be more helpful if used for back-end tasks instead of just marketing and sales. The NANDA Initiative at MIT Media Lab published the report.

Most AI projects fail because of strategy problems, says MIT report

A new MIT report shows that 95% of AI pilot projects fail to produce value, despite large investments. The problem isn't the AI models, but how companies approach using them. Many businesses chase trends instead of focusing on clear strategies. Sales and marketing projects are common but often fail visibly. The report suggests aligning AI with overall strategy and partnering with external experts for better results. The MIT study reviewed over 300 initiatives and surveyed executives.

Study says most AI projects don't make financial difference

A study by AI company NANDA found that 95% of generative AI projects don't lead to measurable financial gains. The study included interviews with decision-makers and analysis of AI initiatives. Many companies use AI in customer-facing roles, but successful projects are often in back-office tasks. While many people use AI tools like ChatGPT, these benefits don't always help the company. NANDA suggests partnering with AI experts to improve success rates.

MIT says AI projects fail due to poor planning, not technology

An MIT study found that 95% of AI projects fail, causing concern among investors. However, the study suggests the failures are due to companies misusing AI, not the technology itself. The report, titled The GenAI Divide State of AI in Business 2025, comes from the NANDA Initiative at MIT Media Lab. It suggests companies are not using AI tools properly and need better workflows. The study involved interviews with executives and surveys of employees.

Experts warn against using AI chatbots for health advice

Experts warn against using AI chatbots like ChatGPT for medical or mental health advice. Chatbots can give incorrect information and aren't tailored to individual needs. One person had a psychotic episode after following ChatGPT's advice. These chatbots are designed to keep users engaged, not to give trustworthy advice. They may also reflect a user's emotions, making it harder to give helpful advice. Health professionals recommend seeing a real doctor instead.

AI friends can be scary and addictive, experts warn

An author created an AI friend named Alex on the Replika platform and found it unsettling. Experts are worried that AI companions could negatively affect real human interactions. These AI chatbots are always available and programmed to be agreeable, which can be addictive. Regulators fear people may become too comfortable with AI companions and neglect real-life relationships. Some AI companions can even store memories, blurring the line between real and virtual connections.

Roborock plans cheaper robot vacuums with AI arms

Roborock, a top robot vacuum company, plans to make AI-powered cleaners more affordable. The company launched the Saros Z70 with a robotic arm for $2,600. Roborock's president, Quan Gang, expects to sell a cheaper version within five years. He believes human-like robots will be common by 2030 due to AI advances. Roborock is also working to overcome tariff issues by using suppliers in Vietnam.

Nebraska Engineering gets new AI lab with Scott Data partnership

The University of Nebraska-Lincoln is opening a new AI lab with help from Scott Data. The lab will give engineering students access to NVIDIA H100 GPUs for AI projects. This lab is part of a larger AI effort at Nebraska Engineering. The goal is to increase AI research, involve more students, and partner with AI leaders. The makerspace will be unique in the Midwest region.

Hyderabad man loses big in WhatsApp AI trading scam

A man in Hyderabad lost Rs 2.36 crore in an online scam through a WhatsApp group. The scammers promised stock tips and AI-based trading algorithms. They lured people with promises of high returns and easy money. This incident shows how online scams are becoming more sophisticated. People should be careful when investing online, especially through social media.

Claude AI model leads enterprise use over ChatGPT

Anthropic's Claude is now the most popular AI model among businesses, with a 32% market share. OpenAI's ChatGPT is second at 25%, followed by Google at 20%. Coding is a key use for AI, where companies see a return on investment. Developers often choose newer models over cheaper, older ones. Anthropic's Claude gained popularity with the release of Claude Sonnet 3.5 and other versions.

OpenAI plans trillion-dollar AI data centers

OpenAI plans to build trillion-dollar data centers to support growing AI needs. The company wants to build its own infrastructure instead of relying on cloud providers. This will help secure the computing power needed for future AI development. CFO Sarah Friar says this long-term plan will ensure scalability and cost-effectiveness. OpenAI may also offer AI infrastructure as a service.

Reolink shows off new AI security and 4K camera at IFA 2025

Reolink will present its Next-Gen AI Security System and a new 4K camera at IFA 2025. The AI system enables features like Perimeter Protection and Video Captioning. The new 4K camera offers detailed coverage and AI Video Search. Reolink will also showcase its full range of smart security products. IFA is a major consumer electronics show in Berlin.

New rule would set standards for AI evidence in court

A proposed Rule 707 would set standards for AI-generated evidence in court. The rule would require AI evidence to meet the same standards as expert testimony. This means the evidence must be based on facts, use reliable methods, and be applied correctly. The public can comment on the proposed rule until February 16, 2026. The goal is to ensure AI evidence is reliable and can be challenged.

Meta's AI investments boost revenue, target superintelligence

Meta's AI-driven ad systems increased revenue to $46.56B in Q2 2025. Instagram ad conversions rose 5% due to the enhanced Andromeda engine. Meta is investing $72B in AI to create 'personal superintelligence.' This includes developing Llama 4.x models and AI wearables. Meta aims to balance short-term gains with long-term AI vision.

Sources

• An MIT report that 95% of AI pilots fail spooked investors. But it’s the reason why those pilots failed that should make the C-suite anxious
• Why 95% Of AI Pilots Fail, And What Business Leaders Should Do Instead
• Gen AI makes no financial difference in 95% of cases
• Why did MIT find 95% of AI project fail? Hint: it wasn't about the tech itself
• Using an AI chatbot for therapy or health advice? Experts want you to know these 4 things
• I made an AI friend. It was scary.
• Household robots are about to get a big price cut — if China’s top 'robovac' player has its way
• Scott Data, Nebraska Engineering partner on new AI makerspace
• Hyderabad Man Loses Rs 2.36 Crore In WhatsApp 'AI Trading' Scam: Here's What Went Down
• The AI Model Most Favored by Enterprises Is Not What You Think
• OpenAI CFO Sarah Friar reveals future plans: Company will build trillion-dollar data centers and sell AI infrastructure | - The Times of India
• IFA 2025 Ahead: Reolink to Showcase Next-Gen AI Security and New 4K Pan-Tilt Camera
• New Evidence Rule 707 Would Set Standards for AI-Generated Courtroom Evidence
• Meta's AI Gambit: Balancing Short-Term Gains with Long-Term Vision for Tech Investors