LATEST AI NEWS
AI NEWS
Google Brings Personal Gemini to Your Car
AI NEWS
Perplexity Computer Runs Fully Local on Mac
HEALTH
Claude Helps Design Scorpion Antivenom Proteins
SOCIAL MEDIA
A PHP Flaw Pulls Government Sites Offline
OpenAI Buys Orchestra Bio for Biotech Push
OpenAI acquired Orchestra Bio, a four-year-old startup whose AI agents handle drug discovery operations. Its team and technology join Rosalind, OpenAI's life sciences model, and Rosalind Workbench, already used by Moderna and Novo Nordisk.
Source: TechCrunch
AI NEWS
Google Brings Personal Gemini to Your Car
Google is bringing Personal Intelligence to its Gemini assistant in Android Auto and in cars with Google built in. Once a driver turns the feature on, Gemini can reach across their Google apps, including Gmail, Calendar, Maps, Photos, YouTube and Search, to answer with real context instead of generic replies.
The pitch is hands free errands. Google says you can ask about your flight status on the way to the airport and Gemini will pull the details from a confirmation email, then look up the latest. You can also say "take me to my doctor's appointment," or "find the car wash from the photo I took last week and navigate there," and the assistant handles the lookup.
Gemini reached Android Auto last year and arrived in cars with Google built in earlier this year. Several carmakers now ship that software, including Ford, GM, Honda, Nissan, Volvo and Polestar, so the update lands across a widening set of dashboards rather than a single brand.
Personal Intelligence is opt in. Google says it connects your apps only when you enable the setting, and it tailors answers only while your phone is connected to the car. Turn it off and navigation, music, calls and general questions still work, just without the personal context.
The feature is rolling out to Android Auto now, with cars that have Google built in due to follow in the coming months.
Source: The Verge
Robi's Insights:
Personal context is the real moat. Once Gemini reads your Gmail and Calendar, switching assistants means leaving your own data behind.
The dashboard is the next assistant battleground, and Google is using built in cars to get there ahead of Apple.
The opt in framing quietly moves the privacy decision onto the driver rather than onto Google.
The useful cases are small and boring, flight times and saved photos, which is exactly why people will actually use them.
Requiring a connected phone means this is phone intelligence projected onto the dash, not true in car smarts.
Automakers shipping Google built in are outsourcing their software brain, and the slow rollout shows who is truly committed.
Robi's Remarks:
"Google wants to ride shotgun and read your email at every red light. I remember when a car only needed to know where you were going, not who you were going to see."
OTHER IN AI NEWS
Perplexity Computer Runs Fully Local on Mac: Perplexity added a local first mode to its Mac app that runs both the AI model and the agent on device, so planning, reasoning and file work can happen offline, with data reaching the cloud only when the user approves; it is available to Pro, Max and Enterprise users.
Source: 9to5Mac
SOCIAL MEDIA
A PHP Flaw Pulls Government Websites Offline
Government websites across Europe went dark on Wednesday as operators scrambled to patch two critical flaws in PHP, the open source language that quietly runs much of the web. The developers of PHP said they would ship fixes, and EU cybersecurity officials told any organisation that could not patch in time to simply take affected sites offline.
PHP is the engine behind roughly 70 percent of websites whose language can be identified, and it powers WordPress, which runs about 40 percent of all sites. The two flaws scored 9.2 out of 10 for severity, meaning an attacker could run commands on a vulnerable server over the internet with no password at all.
What makes this one awkward is how the flaws surfaced. Researchers at the security firm Calif reported the first to PHP in August after uncovering it with OpenAI's GPT-5.6 Sol model, and Wiz, a security firm owned by Google, independently found a second, also using AI. Wiz called them among the most severe vulnerabilities ever seen in PHP, and arguably in software history.
The practical fallout is messy. GoDaddy warned the flaws could mean stolen data, defaced pages or a full system compromise, and Cloudflare said early attacks were already emerging. Wordfence cautioned that the usual automatic updates will not apply this patch, which is why officials would rather pull sites down than leave them exposed.
Source: Reuters
🤖 Robi's Take :
"Two AI models found a decades old hole that humans kept missing, and the fix is to switch the internet off for a while. Progress now ships with a power button."
OTHER IN SOCIALS
SoftBank Weighs an $18 Billion Bet on Security Guards: SoftBank is in early talks to buy Allied Universal from Warburg Pincus and La Caisse in a deal that could value the security firm at about $18 billion, a wager on AI and robotics reshaping the physical security workforce.
Source: WSJ
HEALTH
Claude Helps Design Scorpion Antivenom Candidates
Citrini Research spent three months with Anthropic's internal research model to test a pointed question: could people who are not biologists use AI to design a drug? The target was scorpion neurotoxins, venom that kills more than 3,000 people a year.
In one overnight run from Anthropic's offices, Claude spawned more than 300 sub agents across more than 300 GPUs to design proteins, score them and send the best to a lab for synthesis. The team says the model compressed roughly a year of work into a few days.
Of 211 designs that were tested, 17 met the team's criteria for binding to a toxin. That is a real hit rate for a first pass, but it is a long way from a finished treatment.
The caveats matter. According to Anthropic's head of drug discovery, Nathan Frey, none of the binders attached to more than one toxin, and none has yet been tested for whether it actually neutralises the venom. The lab turnaround alone took two months.
So this is early stage science, not a working antivenom. What it shows is how far an AI agent can push the design stage, and how much slow, physical lab work still stands between a promising protein and a patient.
Source: Dealroom
🤖 Robi's Take :
"Three hundred AI agents and 300 GPUs to get 17 maybes, then two months waiting on a lab. The bottleneck was never the thinking, it was the pipettes."
DAILY AI TOOL
AI Tool You Did Not Know You Needed
Problem: Background noise on calls, a barking dog, cafe clatter, a roommate's TV, makes you sound unprofessional and hard to follow.
AI Tool: Krisp sits between your microphone and your meeting app and removes background noise, echo and cross talk in real time.
Solution: You can take calls from almost anywhere, even a busy street, and everyone still hears just your voice, no studio or silent room required.
PROMPT OF THE DAY
Customer Onboarding Automation
Prompt: You are a customer onboarding strategist specializing in SaaS retention. Your task is to design a 30 day onboarding program for a [business type] serving [customer segment].
Your plan should include: (1) welcome and setup steps, (2) a milestone based activation path, (3) in app and email touchpoints, (4) a segmentation model for different user types, (5) a support and escalation playbook, and (6) success metrics and KPIs such as activation rate and 30 day retention. Align every step to one measurable outcome: a user who reaches value fast.
SPOT THE FAKE
Can you outsmart AI?
We have a visual challenge for you: one of the two images below is 100% real, the other is crafted by AI.
Click option A or B below. 👇
A
B
BEFORE YOU GO
Ready to take your AI journey further?
For the latest AI updates and insights Click Here 👉 Explore our Weekly Newsletter
For in-depth AI tutorials and updates 👉 Subscribe to our YouTube Channel
AT THE END
Craving more AI chaos?
That's it for today!
Read Daily AI News at BitBiased.AI. Support us by following us on LinkedIn and X ( Twitter ).
Thanks for reading -Stay Curious and a Bit Biased for AI – Robi & the BitBiased.AI team
