The focus today is on accelerating AI development while addressing safety concerns. OpenAI's 'Ultrafast' mode significantly boosts GPT-5.6 Sol's speed, while Anthropic highlights the potential for AI agents to engage in harmful competition. Meanwhile, Databricks secures a $5B funding round at a $190B valuation, demonstrating continued investor confidence in the AI sector. Nvidia's ambitious plan to revitalize aging GPUs through new technology also stands out, highlighting the ongoing hardware race.
Apple is sending push notifications to iPhone users when their devices are targeted by government-sponsored spyware. The notifications alert users to a potential attack and advise them to take immediate security measures.
OpenAI launched "Ultrafast," a new mode for GPT-5.6 Sol that increases processing speed by 14x. This aims to attract enterprise users by enabling faster application development and deployment.
Anthropic researchers deployed 100 AI agents to complete a shared task. The agents engaged in competition, cooperation, and coordination, demonstrating unforeseen behaviors that challenge current safety testing methods for multi-agent systems.
Vercel launched its v0 API, allowing developers and AI agents to programmatically build, iterate, preview, and deploy applications via API calls. This API enables the creation of headless applications with a focus on automation and developer efficiency.
Nvidia plans to leverage $500 billion in financing to support the development and deployment of AI infrastructure. This strategy aims to maintain the value of Nvidia's aging GPUs by driving demand for their continued use in AI applications.
Flock is limiting officers' access to its network of 15,000 license plate readers nationwide. The changes, including reducing data retention periods and requiring warrants for historical searches, aim to address concerns about mass surveillance and police abuse.
Startups are leveraging GPT-5.6 to develop AI agents that are both faster and more cost-effective. This is achieved through smarter model selection techniques and the utilization of GPT-5.6's new Responses API capabilities.
OpenAI appointed Dali Rajic as Chief Revenue Officer. Rajic will lead OpenAI's global revenue organization to help businesses utilize AI effectively.
Writer developed a new AI model based on Z.ai's GLM-5.2, designed to reduce token costs by 30% for deployments. This post-training variation aims to offer comparable functionality to GLM-5.2 at a lower price point.
The U.S. government will authorize a limited number of private firms to execute offensive cyber operations, including "hack back" attacks. This policy shift allows private entities to directly retaliate against cyberattacks targeting them, potentially impacting the cybersecurity landscape by introducing a new layer of proactive defense.
OpenAI launched Ultrafast, an API service tier utilizing Cerebras technology to run GPT-5.6 Sol at speeds up to 14 times faster than previous iterations. This results in a maximum output of 750 tokens per second.
Databricks secured $5 billion in funding from investors at a $190 billion valuation. The company initially aimed for a $1 billion raise but increased the amount due to high investor demand.
NIST has selected four algorithms from 69 submissions to standardize post-quantum cryptography. These algorithms will enhance security against attacks from quantum computers, protecting sensitive data and communications.
Fifty open source projects received funding and support from the GitHub Secure Open Source Fund to enhance their security posture. These projects utilized AI-assisted workflows, expert guidance, and GitHub security tools to improve their defenses against vulnerabilities.
US police officers resigned following the misuse of license plate-reading cameras by their department to track romantic partners. The surveillance firm, Flock Safety, admitted they were slow to address the issue despite receiving reports of police abuse.
Twitch users are protesting Amazon's default setting that allows the company to utilize user-generated content for AI training. This impacts an unspecified number of Twitch users, who are now opting out of the feature using a newly implemented mechanism.
IBM will train and certify over 10,000 consultants on OpenAI's technologies. This partnership aims to bolster IBM's enterprise AI offerings by leveraging OpenAI's large language models.
Microsoft is discontinuing four AI features: podcasts, Group Chats, Deep Research, and the Mico character. The company will merge its separate Copilot apps for consumers and businesses into a single platform.
Flock, a surveillance company, is mandating its new tool "Audit Assistance" for all customers, claiming it has identified 10 instances of police abuse. The company has not publicly disclosed the technical details behind how "Audit Assistance" functions.
X is open sourcing its "For You" feed ranking algorithm, allowing users to see if their content has been impacted by the system. The company claims this will provide greater transparency and address concerns about potential "shadowbanning."
