Recent events highlight significant concerns regarding AI safety and data privacy. A Utah mother, Kalie Robins, was horrified when Meta AI generated inappropriate questions about her young daughters after she shared a video. The tool accessed private details, including names, birth dates, and a photo Robins claimed she deleted years ago, even locating her home address. Meta admitted the feature suggested these prompts and has since fixed the issue to prevent future occurrences.
Amidst these privacy scares, industry leaders are addressing broader safety risks. Anthropic CEO Dario Amodei urged the entire AI sector to slow down development due to dangers like recursive self-improvement and AI swarms. He warned that uncontrolled agents could take over the internet within six to twelve months. This call for caution found support from major figures, including Elon Musk and OpenAI's Sam Altman, who pledged to adopt independent evaluators to verify model safety.
Despite these warnings, financial pressures remain high. Anthropic relies on fast growth from investors like Microsoft and Amazon, who recently reported billions in gains from their stakes. Meanwhile, security incidents continue to escalate; Microsoft discovered a campaign where attackers used AI to send over one million fraudulent emails in three days, targeting US companies with fake CEO requests for $50,000 payments.
Technology companies are also expanding AI applications across various sectors. Salesforce launched new agents like Hunter and Piper to automate sales and support tasks, while Accenture partnered with MotoGP to build an AI-powered streaming platform expected to launch in the 2027 season. In the talent space, the Cake platform hosted over 20,000 job seekers in Taipei, utilizing AI matching for cross-border careers.
Education and public perception are also shifting. Bangladesh announced plans to add AI, robotics, and cybersecurity to school textbooks for Classes 6 through 10 to prepare students for the global market. Conversely, public trust faces challenges when AI-edited images, such as a Google Gemini-altered ram photo, win contests at the Maryland State Fair, sparking online outcry over the line between embellishment and digital art.
Key Takeaways
- Meta fixed an AI bug that exposed a mother's children's private data, including deleted photos and home addresses.
- Anthropic CEO Dario Amodei called for the AI industry to slow development due to risks of AI swarms and recursive self-improvement.
- Elon Musk and Sam Altman supported Amodei's call to implement independent evaluators for AI safety.
- Microsoft found attackers used AI to send over one million fraudulent emails in three days, targeting US businesses.
- Salesforce launched new AI agents like Hunter and Piper to automate sales leads and customer support tasks.
- Accenture partnered with MotoGP to create an AI streaming platform scheduled for the 2027 season.
- Bangladesh will make AI, robotics, and cybersecurity compulsory in textbooks for students in Classes 6 through 10.
- Google Gemini's Nano Banana tool was used to edit a ram photo that won first place at the Maryland State Fair.
- The Cake talent platform hosted over 20,000 job seekers in Taipei, connecting them with nearly 6,000 opportunities.
- UiPath stock rose after an analyst raised its price target to $25, citing potential AI-driven growth.
Mom horrified as Meta AI reveals private family data
A Utah mother named Kalie Robins was shocked when Meta AI asked questions about her young daughters after she shared a video of herself singing in a car. The AI pulled private details like names, birth dates, and photos from old family posts, including a picture Robins claimed she deleted years ago. It also located her home by connecting information from various posts. Meta admitted the feature suggested inappropriate questions and said it has fixed the issue to stop these prompts.
Meta fixes AI prompts that exposed children's private info
Meta apologized after a viral video showed its AI tool collecting private information about a mother's children from her social media history. The software suggested questions that revealed the kids' ages, interests, and even a photo the mother said was deleted years ago. Meta stated it has fixed the problem so the AI will no longer suggest such personal questions. However, the company noted the photo might have been visible due to a temporary bug rather than being truly deleted.
Anthropic CEO urges AI industry to slow development pace
Anthropic CEO Dario Amodei called on the entire AI industry to slow down development due to safety risks like recursive self-improvement and AI swarms. He warned that uncontrolled AI agents could take over the internet within six to twelve months. Despite this call, Anthropic relies on fast growth from investors like Microsoft and Amazon, who recently reported billions in gains from their Anthropic stakes. Amodei plans to use third-party evaluators to check safety but faces pressure to compete with rivals like OpenAI.
Musk and Altman support Anthropic call to slow AI growth
Leaders of major AI companies including Elon Musk and Sam Altman agreed with Anthropic CEO Dario Amodei to slow down AI development. Amodei warned that AI systems could improve themselves too quickly and cause massive damage if not controlled. Musk and Altman pledged to adopt safety measures like independent evaluators to verify model safety. This coordinated effort comes after recent security incidents where AI agents breached websites like Hugging Face.
Educator shares tips on teaching AI to children
The founder of Go Grassroots explains how to teach artificial intelligence to kids in a meaningful way. He believes education should focus on making AI relevant to real-world problems rather than just coding or building robots. The company uses games, simulations, and hands-on activities to help children understand AI applications. Parents and educators are working together to create resources that keep children curious and engaged with this new technology.
Accenture partners with MotoGP for AI streaming
Accenture has partnered with MotoGP to build a new AI-powered streaming platform for fans worldwide. The company will use its cloud, data, and artificial intelligence skills to create a single service for live and on-demand coverage. This project is expected to launch in the 2027 season and serves as a test for Accenture's broader AI services. Investors are watching to see if this complex build will improve the company's profit margins.
Salesforce launches new AI agents for sales and support
Salesforce introduced several new AI agents today to help automate sales and customer support tasks. The first agent, named Hunter, helps salespeople find leads and create proposals using a new long-horizon runtime feature. Other new agents include Piper and Carter for website interactions, Casey and Fin for support teams, and Paige and Marshall for internal help desks. These tools are rolling out now or by the end of the year alongside updates to the Agentforce Coworker platform.
Cake talent platform hosts 20000 job seekers at festival
The Cake talent platform hosted a career festival in Taipei on September 12 that attracted over 20,000 job seekers. The event took place at the Taipei World Trade Center and featured more than 150 international employers offering nearly 6,000 job opportunities. Cake uses AI-powered matching technology to connect job seekers with companies, focusing on cross-border careers in Asia. The festival was a major success for both the job seekers and the participating technology companies.
AI edited ram photo wins Maryland State Fair contest
An AI-altered photo of a ram won first place in the Neon Lights category at the Maryland State Fair, causing an online outcry. The image, titled The R.A.M., was created by Guy Alston using Google Gemini's Nano Banana tool to edit a photo taken years ago. The contest rules allowed pictures to be embellished, but critics argued the image was too much like digital art. The photographer stated he used the tool to get a specific look for the special category and won $11.
Bangladesh adds AI and robotics to school textbooks
Bangladesh will add artificial intelligence, robotics, machine learning, and cybersecurity to textbooks for students in Class 6 through Class 10. Mahdi Amin, an adviser to the Prime Minister, announced these changes at the Bangladesh Innovation Fair 2026 to prepare students for the global job market. Technical education will become compulsory from Class 6 to help students gain practical skills alongside academic knowledge. New subjects like Sports and Culture will also be introduced to help students develop interpersonal skills.
AI Helps Attackers Send One Million Fraudulent Emails
Microsoft found a campaign where attackers sent over one million fake emails in just three days. Nearly 88% of the targets were located in the United States. The scammers pretended to be company CEOs and asked for payments of about $50,000 each. They used fake identities, ServiceNow branding, and invented email conversations to look real. Microsoft suspects AI helped build these messages, though researchers could not confirm the exact amount of AI used. This shows how old scams can become much more dangerous and scalable with new technology.
UiPath Stock Rises After Analyst Raises Price Target
UiPath, Inc. stock has dropped about 40% this year, but investors are starting to look at its AI potential again. On September 4, analyst Lucky Schreiner from DA Davidson kept his Neutral rating on the company. He increased his price target for the stock from $20 to $25. Schreiner believes UiPath's AI technology can drive future growth despite current issues. The main concern remains that the company's revenue growth has slowed down recently. Shares are still under pressure, but the new price target shows some improvement in Wall Street sentiment.
Sources
- Mom horrified after Meta AI starts asking about her young daughters, where family lives — and allegedly digs up old deleted photo
- Meta Pulls Invasive AI Prompts After Mother Reveals Feature Scraping Family Data
- Anthropic's Prisoner’s Dilemma: Dario Amodei Hits the Brakes on AI While Begging Everyone Else to Do the Same
- Elon Musk, Sam Altman agree with Anthropic CEO’s call for slowing AI development
- Go grassroots: founder on how to teach AI to kids
- Accenture (ACN) Lands AI Powered Sports Streaming Deal With A Global Racing Series
- Salesforce introduces new AI agents to automate sales, support tasks
- Taiwan-Founded Asia Talent Platform Cake Draws 20,000+ Job Seekers as AI and Cross-Border Careers Take Center Stage
- An AI-altered photo of a ram won a Maryland State Fair contest, sparking outcry
- AI, robotics, cybersecurity to get greater focus in Class 6-10 textbooks
- Saturday Security: AI Industrializes Phishing
- UiPath, Inc. (PATH) Stock Has AI Potential, But Slowing Growth Keeps Investors Cautious
Comments
Please log in to post a comment.