AI news and funding updates from the last 24 hours(19th April 2025):

OpenAI

Memory with Search OpenAI’s new “Memory with Search” option allows ChatGPT to pull in remembered details when crafting web‐search queries—refining prompts with context from past conversations to deliver more relevant results. Users who prefer traditional searches can turn this feature off in settings.

Name‑by‑Name Personalization Recently, ChatGPT has sometimes referred to users by name during its internal reasoning, despite never being explicitly told their names. Many users have described this unexpected personalization as “creepy” and have called for an explanation from OpenAI.

Increased Hallucinations in o3 & o4‑mini Despite being state‑of‑the‑art in coding and math, the new o3 and o4‑mini reasoning models have shown an uptick in producing fabricated or “hallucinated” outputs compared to previous generations. OpenAI acknowledges the regression and is investigating ways to improve factual accuracy.

Mechanize

Startup Mission Mechanize, launched by AI researcher Tamay Besiroglu (of Epoch), has a bold goal of achieving “full automation of all work” through AI agents. The announcement sparked debate over job displacement and the impartiality of Besiroglu’s research institute. Mechanize admits current AI agent limitations and is initially focusing on automating white‑collar tasks, with a long‑term vision of economic abundance through automation.

Anduril

AI Weapons Race Palmer Luckey, founder of defense‑tech firm Anduril, argues the U.S. has already crossed the “point of no return” in autonomous weaponry and risks falling behind China without increased investment. He cites scenarios—such as a hypothetical Chinese invasion of Taiwan—where U.S. forces might lack the autonomous capabilities needed for an effective response, and urges policymakers to “double down” on AI weapon development.

Other News

Indian GenAI Adoption

Indian enterprises are accelerating their adoption of Generative AI: spending is projected to grow at 35% annually, reaching $9.2 billion by 2028, driven by the promise of significant productivity gains despite challenges around data quality.

U.S. AI‑Powered Immigration Surveillance

Under the Trump administration, agencies like DHS and CBP have expanded use of AI tools—such as facial recognition and social‑media monitoring—to track and arrest immigrants. Critics warn that a focus on deportation numbers over accuracy has led to wrongful arrests and broader privacy concerns.

India’s ‘Adipoli’ AI Server

India has unveiled ‘Adipoli’, its first fully indigenously designed AI server. Equipped with eight GPUs and developed under the Make in India initiative, it symbolizes the country’s push toward self‑sufficiency in AI infrastructure and reduced reliance on foreign technology.

Listen to us here:

👉🏻 Spotify:

https://open.spotify.com/show/3SxK3fD2xqV5iLW48VFOxI?si=8e1fb9738b2342d6

👉🏻 Amazon Music:

https://music.amazon.in/podcasts/396d7f14-97ac-4f8a-b2bd-89b34d84cfdb/ai-news-and-funding-updates

👉🏻 YouTube:

https://youtu.be/B3o6wrqMLME