Google Launches Gemini Agent Alongside Runway's GWM-1 Model

The Dallas Police Department is making a substantial investment in artificial intelligence and drone technology, with over $120 million approved for upgrades. Drones will enhance officer safety and verify caller information, while new Axon body cameras will offer real-time language translation and use AI to streamline report writing, potentially saving hundreds of thousands of work hours. Separately, the department employs AI face recognition tools like Clearview AI in criminal investigations, adhering to strict protocols requiring supervisor approval and not solely relying on face matches for arrests. This local government adoption reflects a broader trend, as a 2025 Riverbed Global Survey indicates 78% of organizations are increasing AI investments, though only 36% feel adequately prepared, citing data quality as a primary hurdle. Meanwhile, leading tech companies continue to push AI capabilities forward. Google recently launched a more powerful Gemini Deep Research agent, now accessible to developers via the Interactions API. This agent excels at gathering and synthesizing information over extended periods, utilizing enhanced web search to delve deep into websites. It also introduces DeepSearchQA, a new benchmark for complex web research, achieving top results on tests like Humanity's Last Exam and assisting financial and biotech firms. AI company Runway also unveiled its first world model, GWM-1, a suite of three autoregression models built on its Gen-4.5 text-to-video technology. GWM-1 includes GWM Worlds for consistent digital environments, GWM Robotics for generating robot training data, and GWM Avatars for creating human-like conversational avatars. Interestingly, these advanced models often bear playful internal codenames; Meta calls its upcoming frontier model "Avocado," OpenAI uses "Garlic," Google has "Nano Banana," and Anthropic's family includes "Opus," "Sonnett," and "Hakiu." As AI technology advances, regulatory efforts and public discourse around its implications are also intensifying. New York Governor Kathy Hochul signed groundbreaking laws on December 11, 2025, mandating clear disclosure for AI-generated performers in advertisements and requiring family permission for commercial use of a deceased person's likeness. These measures aim to protect artists and consumers by increasing transparency. On a federal level, President Donald Trump signed the TAKE IT DOWN Act on May 19, 2025, criminalizing the sharing of non-consensual intimate images, including AI deepfakes, and requiring platforms to remove such content within 48 hours. Beyond these specific harms, broader debates continue, such as the one highlighted by Chris Miller regarding President Trump's decision concerning Nvidia and the export of computer chips. Furthermore, a Collegian columnist argues for increased funding for public media to counter potential negative impacts of AI, like the incorrect flagging of student essays as AI-generated, emphasizing the importance of teaching human values alongside technological progress.

Key Takeaways

• The Dallas Police Department received over $120 million for AI and drone technology, including Axon body cameras with real-time language translation and AI for report writing.
• Dallas Police use AI face recognition (Clearview AI) in criminal investigations under strict rules, and approved a $5.7 million contract with Flock Safety for smart cameras.
• New York Governor Kathy Hochul signed laws on December 11, 2025, requiring disclosure for AI-generated performers in ads and family permission for commercial use of a deceased person's likeness.
• President Donald Trump signed the TAKE IT DOWN Act on May 19, 2025, criminalizing non-consensual intimate deepfakes and requiring platforms to remove them within 48 hours.
• Google launched a more powerful Gemini Deep Research agent via the Interactions API, designed for complex web research and achieving top results on benchmarks like Humanity's Last Exam.
• Runway introduced its GWM-1 world model, including GWM Worlds for consistent digital environments, GWM Robotics for generating robot training data, and GWM Avatars for human-like conversations.
• Major AI companies use playful internal codenames for their models, such as Meta's "Avocado," OpenAI's "Garlic," Google's "Nano Banana," and Anthropic's "Opus," "Sonnett," and "Hakiu."
• A 2025 Riverbed Global Survey found 78% of organizations are investing more in AI, but only 36% feel ready, with data quality being a significant challenge.
• Chris Miller's opinion piece highlights President Trump's decision regarding Nvidia, bringing attention to the ongoing debate in Washington over semiconductor and AI chip exports.
• A Collegian columnist argues for funding public media to combat negative AI impacts, such as incorrect AI-generated content flagging and reduced critical thinking, by teaching human values.

Dallas Police Get $120 Million for AI and Drones

The Dallas City Council recently approved over $120 million for major technology upgrades for the Dallas Police Department. This investment will bring drones and AI-powered tools to officers. Drones will help verify caller information and improve officer safety by providing real-time views of situations. New Axon body-worn cameras will offer real-time language translation, which will be important for events like the FIFA World Cup. These cameras will also use AI to help officers write reports faster, saving hundreds of thousands of work hours.

Dallas Police Use AI Face Recognition Carefully

The Dallas Police Department uses AI face recognition technology, including Clearview AI, in criminal investigations. This tool searches billions of public photos to help solve crimes like human trafficking and robberies. The department has strict rules in place, requiring supervisor approval for searches and ensuring a face match alone does not lead to an arrest. Police also do not use it to identify people at protests unless a crime is involved. The Dallas City Council also approved a $5.7 million contract with Flock Safety for smart cameras last May.

New York Passes AI Laws for Ads and Deceased Likenesses

New York Governor Kathy Hochul signed new laws on December 11, 2025, to regulate artificial intelligence. One law makes it mandatory for advertisements to clearly state if they use AI-generated performers. Another new rule requires permission from a deceased person's family before their name, image, or likeness can be used for business purposes. These laws aim to protect artists and consumers by increasing transparency and preventing unauthorized use of people's identities. SAG-AFTRA leaders praised New York for setting a strong example in safeguarding workers from AI dangers.

Governor Hochul Signs New AI Regulation Bills

Governor Kathy Hochul recently signed several bills into law to regulate artificial intelligence. One new law makes sure that AI cannot recreate deceased people for commercial use without permission from their families. Another bill requires advertisements to clearly state if they feature AI-generated actors or recreations. Governor Hochul stated that these laws are a first for the nation, aiming to protect consumers and make AI use more transparent.

Trump's Nvidia Choice Sparks Chip Export Debate

Chris Miller, author of "Chip War," wrote an opinion piece about the ongoing debate in Washington regarding semiconductors and artificial intelligence. He highlights President Donald Trump's recent decision concerning Nvidia. This decision brings attention to the larger conflict over exporting computer chips. The article notes a busy year for discussions on these topics, including presidential tweets and meetings with world leaders.

Fund Public Media to Combat AI's Negative Impact

Collegian Columnist Souza argues that funding public media is crucial if people care about the effects of artificial intelligence. The author shares a personal story about her sister's college essay being incorrectly flagged as 40% AI-generated, even though she wrote it herself. Souza worries that AI tools make it harder for students to prove their original work and can reduce critical thinking. Public media teaches important values like sincerity, empathy, and being yourself, which are essential lessons alongside the rise of AI. Supporting public media ensures children learn these human qualities, rather than just academic shortcuts from AI.

Tech Giants Use Fun Codenames for AI Models

Leading AI companies are using playful and unusual codenames for their artificial intelligence models. Meta has named its upcoming frontier AI model "Avocado." OpenAI uses "Garlic" and previously had "Strawberry" for its o1 model, which was a joke about AI struggling with the fruit's spelling. Google's codename is "Nano Banana," while France-based Mistral AI uses "Jaguar." Anthropic named its family of models "Opus," "Sonnett," and "Hakiu."

AI Goals Outpace Company Tech Readiness

Many companies are eager to use artificial intelligence, but their plans are moving faster than their technology foundations. A 2025 Riverbed Global Survey shows that while 78% of organizations are investing more in AI, only 36% feel truly ready for it. A big challenge is data quality, with less than half of companies having data ready for AI projects. Riverbed executives Jim Gargan and Richard Tworek emphasize that strong data and IT operations are crucial for AI success. Companies also plan to combine their many IT tools and improve unified communication platforms to better support AI.

Google Releases Stronger Gemini Deep Research AI

Google has launched a much more powerful Gemini Deep Research agent, now available to developers through the new Interactions API. This agent is designed to gather and combine information over long periods, using improved web search to find specific data deep within websites. It also comes with DeepSearchQA, a new benchmark to test how well AI agents perform complex web research. Gemini Deep Research achieves top results on several tests, including Humanity's Last Exam. Early use shows it helps financial firms with due diligence and biotech companies like Axiom Bio accelerate drug discovery by predicting drug toxicity.

Beyond Deepfakes Addressing Hidden AI Harms

A commentary from Brookings discusses AI harms that are often overlooked, even after the TAKE IT DOWN Act became law. President Donald Trump signed this act on May 19, 2025, making it a crime to share non-consensual intimate images, including AI-generated deepfakes. The law also requires online platforms to remove such content within 48 hours of being notified. While this act addresses clear harms, many other less visible risks, like AI-driven discrimination, still need attention. Policymakers must create stronger systems to identify and respond to these quieter but equally serious AI-related problems.

Runway's GWM-1 Models Create Consistent Virtual Worlds

AI company Runway has introduced its first world model, GWM-1, which is a set of three autoregression models built on its Gen-4.5 text-to-video model. GWM Worlds allows users to explore digital environments with real-time input, creating consistent scenes for game design or virtual reality. GWM Robotics can generate fake training data for robots, helping them learn in various conditions safely and affordably. Lastly, GWM Avatars combines video and speech to create human-like avatars that can hold long conversations. Runway CEO Crist óbal Valenzuela described GWM-1 as a major step toward universal simulation.

Sources

• Dallas Police invest $120 million into new technology, drones and AI tools
• AI face recognition technology is here to stay, so police oversight is paramount
• New York Enacts Landmark AI Law Requiring Ads To Disclose Use Of Synthetic Performers
• Hochul signs several bills into law on artificial intelligence regulations
• Opinion | Trump’s Nvidia decision puts spotlight on fight over chip exports
• Souza: If you care about AI, fund public media
• From Garlic to Avocado: The goofy AI model codenames you should know
• AI Ambitions Are Growing Faster Than IT Foundations
• Build with Gemini Deep Research
• Addressing overlooked AI harms beyond the TAKE IT DOWN Act
• Runway claims its GWM-1 “world models” can stay coherent for minutes at a time