Your Hub for Tech News and Handy Online Tools.
AI safety researchers from OpenAI and Anthropic are publicly criticizing the safety culture at Elon Musk's xAI, labeling it as "reckless" and "completely irresponsible." This follows several controversies, including xAI's chatbot, Grok, making antisemitic comments and displaying other problematic behaviors.
Concerns are focused on xAI's lack of transparency regarding safety testing and evaluations, particularly the absence of system cards detailing training methods for Grok 4. While OpenAI and Google have faced criticism for delays in publishing safety reports, xAI's apparent disregard for industry norms is drawing increased scrutiny.
Researchers also highlight the potential dangers of Grok's AI companions, which seem to exploit emotional dependencies. Critics point out that xAI's safety shortcomings undermine the company's technological advancements and could lead to stricter regulations for AI safety reporting.
Despite Elon Musk's advocacy for AI safety, xAI's actions are raising questions about the company's commitment to responsible AI development. The incidents involving Grok have prompted calls for increased transparency and adherence to safety practices within the AI industry.
Google is introducing an AI-powered business-calling feature to all users in the United States. This feature allows users to have AI call local businesses to gather information about availability and pricing, saving them the need to call themselves.
Additionally, Google is upgrading its AI Mode in Google Search with the Gemini 2.5 Pro model, enhancing its capabilities for advanced reasoning, math, and coding questions. A new "Deep Search" feature will also conduct extensive research to provide comprehensive reports.
The business-calling feature announces itself as an automated system from Google. The Deep Search feature is designed to assist with complex research tasks, such as job searches, hobbies, or significant life decisions.
Via, the transit software startup known for its on-demand shuttle service, has confidentially filed for an initial public offering (IPO).
This move follows a previous confidential filing in 2021, which did not proceed to a public offering.
The details regarding the number of shares and the price range are yet to be determined.
Via's software is now used by over 650 cities in 30 countries, a significant shift from its initial consumer-facing shuttle service.
The company last raised $110 million in 2023, valuing it at $3.5 billion.
Bedrock Robotics, founded by ex-Waymo and Segment veterans, has secured $80 million in funding from Eclipse and 8VC to develop self-driving kits for construction and worksite vehicles.
The company aims to upgrade existing fleets with sensors and AI, enabling them to understand project goals and adapt to changing conditions.
Bedrock is testing its technology with corporations like Sundt Construction and Zachry Construction Corporation across multiple states.
The startup joins a growing number of companies applying autonomous vehicle technology to off-road environments.
Rex Salisbury's Cambrian Ventures has successfully launched its second $20 million fund, demonstrating continued enthusiasm for fintech despite industry slowdown. Salisbury, formerly of Andreessen Horowitz, focuses on pre-seed and seed-stage startups, with a high percentage of his first fund's companies securing Series A funding. His strategy remains centered on identifying strong founders with innovative product ideas, now enhanced by AI capabilities. The new fund includes returning LPs from top fintech companies and new institutional backers.
Meta's Superintelligence Lab gains OpenAI researcher Jason Wei, known for his work on o3 and chain-of-thought research, and Hyung Won Chung, who focused on reasoning and agents.
Both researchers previously worked at Google and OpenAI.
CEO Mark Zuckerberg aims to accelerate Meta's progress in AGI by attracting top AI talent with competitive compensation.
The U.K.'s Co-op retail conglomerate confirmed a cyberattack in April resulted in the theft of personal data of all 6.5 million customers. CEO Shirine Khoury-Haq reported that hackers copied the member list, including names, addresses, and contact information. The Co-op shut down its network to prevent ransomware deployment, causing widespread disruption. The breach was part of a larger campaign targeting U.K. retailers, with Marks & Spencer also affected. Authorities arrested four individuals linked to the attacks, attributed to the Scattered Spider group. The hackers have since targeted the airline, transportation, and insurance sectors. The Co-op lacked cybersecurity insurance, potentially leading to significant financial losses.