Meta is expanding privacy options for its AI services with WhatsApp's new Incognito Chat feature. Available now in the US, UK, and India, this mode uses end-to-end encryption to ensure conversations remain private, even from Meta itself. Chats are not saved on servers and delete immediately after reading, with plans to roll out to more countries soon.
While Meta focuses on user privacy, Anthropic and OpenAI are fueling a surge in AI-driven cyberattacks. Palo Alto Networks warns these threats will become standard within three to five months as advanced models help hackers find software vulnerabilities. Security experts urge businesses to upgrade defenses immediately, and Palo Alto plans to release new virtual patching tools to counter these sophisticated attacks.
Concerns about AI's physical and environmental impact are also growing. Experts warn of new threats to satellite remote sensing systems, including adversarial patches and backdoor attacks targeting the A-train constellation. Meanwhile, insurance companies face challenges as physical AI robots enter the workforce, requiring updated policies to cover potential bodily injuries or property damage caused by malfunctions.
Infrastructure and sustainability remain critical issues. Citrus County, Florida, is considering a one-year moratorium on new AI data center approvals due to fears over power grid strain and water supply depletion. On the environmental front, researcher Sasha Luccioni highlights that AI could produce 3.2 gigatons of CO2 by 2025, urging the industry to adopt more efficient algorithms and transparent data practices.
Innovation continues to drive solutions across sectors. Mantis Biotech combines large language models with physics simulations to create synthetic medical data, helping an NBA team predict athlete injuries and assisting pharma companies in drug trials. Additionally, researchers have developed a new neuromorphic chip using molybdenum disulfide materials that significantly cuts AI energy use through a signal-folding strategy.
Investment strategies are also evolving. Investors in AI biotech are shifting from hype to demanding proof of value, requiring companies to demonstrate clear strategies for solving specific medical problems rather than relying on broad platform claims. This maturing industry now prioritizes asset-centric companies with defined disease targets and realistic regulatory paths.
Key Takeaways
['WhatsApp launches Incognito Chat for Meta AI, offering end-to-end encrypted conversations that delete immediately and are inaccessible to Meta.', 'Palo Alto Networks predicts AI-driven cyberattacks will become the norm within three to five months, aided by new models from Anthropic and OpenAI.', 'Researchers identify new AI threats to satellite remote sensing, including adversarial patches and backdoor attacks targeting the A-train constellation.', 'Sasha Luccioni warns AI could produce 3.2 gigatons of CO2 by 2025, calling for better emissions data and efficient algorithms.', 'Mantis Biotech uses physics simulations and large language models to create synthetic medical data for drug discovery and injury prediction.', 'Insurance companies must adapt policies to cover risks from physical AI robots, such as bodily injury and property damage from malfunctions.', 'Citrus County, Florida, considers a one-year moratorium on AI data center approvals due to concerns over power grid and water strain.', 'Investors in AI biotech now demand tangible value and biological validation, moving away from hype toward asset-centric companies.', 'A new neuromorphic chip using molybdenum disulfide reduces AI energy consumption through a signal-folding strategy.', "WIRED hosts a livestream AMA on May 27 to discuss AI's impact on jobs, addressing concerns about automation and new career paths."]WhatsApp launches private Incognito Chat for Meta AI
WhatsApp has launched a new feature called Incognito Chat that allows users to talk privately with its AI chatbot. This mode uses existing end-to-end encryption to ensure that no one, including Meta, can access the conversations. The chats are not saved on servers and are deleted immediately after being read. Users in the United States, the United Kingdom, and India can use this feature right now. WhatsApp plans to expand the feature to more countries in the coming months.
WIRED hosts livestream AMA on AI and the future of work
WIRED is hosting a live question and answer session on May 27 to discuss how artificial intelligence is changing jobs. A panel of experts will talk about the impact of AI on the job market and answer questions from the audience. Viewers can submit their questions about job security and new opportunities before the event. The session aims to address concerns about robots replacing workers and explore new career paths created by AI.
Experts warn of AI threats to satellite remote sensing
Researchers are studying new dangers where artificial intelligence is used to attack satellite image analysis systems. These counter-AI threats include methods like adversarial patches that can hide objects from detection or backdoor attacks that trick software. The study highlights risks to Earth observation satellites known as the A-train constellation. Experts are working on defense strategies to protect these critical systems from malicious AI interference.
Making AI sustainable requires better data and efficiency
AI researcher Sasha Luccioni argues that the industry needs better emissions data to understand its environmental impact. Current studies predict AI could produce 3.2 gigatons of CO2 by 2025, which equals the emissions of 700 million cars. To fix this, companies must develop more efficient algorithms and sustainable data centers. Experts also suggest reducing unnecessary data collection and creating more transparent AI systems. A holistic approach covering the entire lifecycle of AI is necessary for true sustainability.
Mantis Biotech uses physics to create synthetic medical data
Mantis Biotech is building human digital twins by combining large language models with physics simulation engines. This technology creates scientifically accurate synthetic data for drug discovery, surgical robotics, and sports injury prediction. The company solves the problem of scarce real-world medical data by simulating human anatomy and physiology based on physics laws. An NBA team is already using their system to predict athlete injuries, and they are working with pharma companies for drug trials.
Insurance companies must adapt to risks of physical AI robots
As AI robots move into the physical world to cook food and operate machinery, new insurance risks are emerging. These robots can cause bodily injury, property damage, or business interruptions if they malfunction. Traditional insurance policies like commercial general liability may cover these accidents, but legal definitions of an occurrence are being tested. Courts will need to decide if AI errors count as unexpected accidents or intentional conduct. Experts say organizations should review their insurance coverage as physical AI becomes more common.
Palo Alto warns AI cyberattacks will become normal soon
Palo Alto Networks predicts that AI-driven cyberattacks will become the new norm within three to five months. New advanced AI models from companies like Anthropic and OpenAI are making it easier for hackers to find software vulnerabilities. Security experts say businesses must upgrade their defenses immediately to stay ahead of these threats. Palo Alto plans to release new virtual patching tools very soon to help protect against these sophisticated attacks.
Citrus County considers pausing AI data center approvals
Citrus County officials in Florida are considering a one-year moratorium on new AI data center applications. Residents are concerned about the strain on the power grid, water supplies, and local property values. The proposed pause would stop zoning changes for data centers at the Holder Industrial Park. This decision comes after the state gave local governments the final say on data center planning. Officials want time to research the impacts and draft better rules to protect the community.
Investors now demand proof of value from AI biotech firms
Investors in AI biotech are shifting from hype to a focus on tangible value and biological validation. Companies must now show clear strategies for solving specific medical problems rather than just promising fast drug discovery. Experts describe this change as a move toward asset-centric companies with defined disease targets. Due diligence has increased, requiring proof of sound science and realistic regulatory paths. The industry is maturing as investors look for practical applications instead of broad platform claims.
New chip design cuts AI energy use with signal folding
Researchers have developed a new neuromorphic chip that significantly reduces the energy needed to run AI models. The device uses a signal-folding strategy to lower operating voltage and improve precision during calculations. Built with molybdenum disulfide materials, the chip mimics the structure of the human brain for efficiency. Tests show it performs vector-matrix multiplications with high accuracy while using much less power than current systems.
Sources
- WhatsApp Adds Meta AI Chats That Are Built to Be Fully Private
- Submit Your Questions: AI Is Changing Your Job—Now What?
- The emerging AI battlespace: Counter-AI threats to AI-powered satellite remote sensing analysis
- What It Will Take to Make AI Sustainable
- Claude's Corner: Mantis Biotech — The Digital Twin Factory Solving Medicine's Data Problem With Physics
- "Yes, AI chef!": A recipe for insuring physical artificial intelligence
- AI-driven cyberattacks will start to be the 'new norm' in months, Palo Alto warns
- Citrus County Commission considers pause on AI data center applications
- What investors now expect from AI biotech companies
- Signal-folding design helps neuromorphic chip slash AI energy use
Comments
Please log in to post a comment.