LATEST AI NEWS
AI NEWS
SpaceXAI Launches Grok Bot
AI NEWS
Gemini Reaches 1 Billion Users
HEALTH
China Frees an AI Tool to Decode Rare Diseases
SOCIAL MEDIA
An Anthropic Model Chips Away at the Riemann Hypothesis
ChatGPT Comes to Linux
OpenAI released a dedicated ChatGPT desktop app for Linux this week, closing the last major gap in its desktop lineup. The preview supports Ubuntu, Debian, and Fedora, and brings ChatGPT, ChatGPT Work, and Codex to open-source developers worldwide.
Source: TechCrunch
AI NEWS
SpaceXAI Launches Grok Bot
SpaceXAI has launched Grok Bot, an early beta that turns its Grok models into a team of always-on AI agents you can hand real work to. Each Bot runs on its own computer in the cloud, signs into the tools your team already uses, and keeps working around the clock, checking back only when something needs your approval.
Rather than building rigid workflows first, you message a Bot the way you would text a colleague and describe the task. It works across apps, inboxes, and websites, including platforms that have no clean API or integration, and returns when the job is finished. Bots remember past conversations and pick up the same thread later on desktop or iOS.
Users can run several Bots in parallel, with one acting as a chief of staff that coordinates specialists for inbox management, expenses, recruiting, bug fixes, or operations. Bots can message each other, share context, and hand off work inside group chats, pulling a human in only for judgment calls. Show a Bot how you do a job once and it saves the steps as a routine it can repeat.
Inside SpaceXAI, teams already use Bots to update the CRM from call transcripts, seat new hires, process invoices from Gmail, and reproduce bugs before filing tickets. The pitch is less about prompting and more about delegation, which is where many companies still struggle to get value from AI.
Grok Bot is in beta and available today for SuperGrok Heavy, Cursor Ultra, and Cursor Teams Premium subscribers on desktop and iOS. Enterprise users can join a waitlist. As an early beta handed broad access to your live tools, it is worth watching how much oversight each Bot still needs.
Source: SpaceXAI
Robi's Insights:
Delegation, not prompting, is the real pitch here, and that is a harder habit for teams to build than any feature.
Giving a Bot its own logins means your security team now owns a new category of non-human accounts to monitor.
The 90 to 100 percent gap SpaceXAI cites is exactly where most pilots quietly stall, so finishing tasks is the claim to test.
Beta access limited to premium tiers keeps the blast radius small while the coordination logic gets stress tested.
Bots messaging each other sounds efficient until two of them disagree about who owns a task.
Watching a Bot learn your workflow once is convenient, but it also bakes in whatever bad habits you demonstrate.
Robi's Remarks:
"An agent with its own computer and all your passwords is either the best hire you made this quarter or an incident report waiting to happen. The difference is entirely in how much you check its work."
OTHER IN AI NEWS
Gemini Reaches 1 Billion Monthly Users: Google CEO Sundar Pichai said the Gemini app has passed 1 billion monthly active users, its 14th product to hit that mark, with 63 percent of users talking to the assistant by voice and more than 150 million images generated daily.
Source: TechCrunch
SOCIAL MEDIA
An Anthropic Model Made Real Progress on the Riemann Hypothesis
Anthropic said on Monday that an as-yet-unreleased model made significant progress on the Riemann hypothesis, a 150-year-old question about the distribution of prime numbers that carries a $1 million bounty for a general proof. The model did not solve it, but it meaningfully raised the lower bound of solutions for which the hypothesis holds true.
The way the result was produced is the striking part. An Anthropic staff member with no serious mathematical training asked the model to take a real stab at the problem, then let it coordinate on its own for about a day and a half. The model tested 650 different ideas across 60 subagents and spent 31 million output tokens in total.
According to a footnote in the paper, two of the 60 subagents developed the key ideas, 13 fed ideas to them, 30 tried and failed to find new ones, 13 checked the arguments, and two wrote the initial draft. Two in-house mathematicians confirmed the finding, which was formalized using the open-source proof assistant Lean.
It is the latest in a run of machine-assisted math results this year, and it is reopening a hard question for the field: if theorems stop being tied to named authors, what happens to credit and responsibility? A group of mathematicians raised exactly that concern in a June declaration, while Fields Medal winner Timothy Gowers suggested the shift might not be all bad.
Source: TechCrunch
🤖 Robi's Take :
"Sixty subagents, 31 million tokens, and a day and a half to nudge one bound on a 150-year-old problem. Impressive, yes, but nobody should confuse a very expensive lower bound with the million dollar proof."
OTHER IN SOCIALS
Saber Denies Swapping a Writer for ChatGPT: Saber Interactive said it did not replace a writer with ChatGPT on its driving game Rideshare Stimulator, but acknowledged it will use AI for in-game passenger dialogue because the number of passengers is effectively infinite, even though the game's Steam page carries no AI disclosure.
Source: PC Gamer
HEALTH
China Releases a Free AI Tool to Decode Rare Diseases
Chinese researchers have released an open-source AI system built to shorten the long and often fruitless search for a rare disease diagnosis. Most rare conditions are genetic, and patients can spend years visiting specialists while the cause stays hidden inside three billion base pairs of DNA.
The system, called OneGenome, comes from BGI-Research. According to the team, it does not simply read DNA sequences. It is trained to reason over them, combining a genomic foundation model named Genos with large language model capabilities to interpret the clinical consequences of gene mutations. The team notes that genomic sequences are connected rather than independent, so removing one can change the entire network of interactions.
The team says OneGenome outperformed general models such as DeepSeek-v4 and traditional gene models across several diagnostic and medication-guidance tests. Automating the screening and reasoning steps compresses a process that used to take years into a few hours, while producing a traceable analysis a doctor can check.
The group also applied the model to rare cancers. In one late-stage case of a rare soft-tissue tumor, it flagged 11 therapeutic targets that standard testing had missed. Chief scientist Liu Shiping framed the tool as a global public good, especially for countries with limited medical resources, and stressed that AI is meant to give doctors leads, not final diagnoses.
Source: South China Morning Post
🤖 Robi's Take :
"Free and open-source is the part that actually matters here. A diagnostic tool locked behind a paywall helps nobody in the places where the diagnostic odyssey is longest and the specialists are fewest."
DAILY AI TOOL
AI Tool You Did Not Know You Needed
Problem: You explain a process clearly in writing, but the wall of text still loses the room and nobody remembers the flow.
AI Tool: Napkin AI reads text you paste or import and turns it into editable diagrams, flowcharts, mind maps, and infographics, no prompting or design skill required.
Solution: Drop in your notes and export a clean visual to PowerPoint, Google Slides, or a doc in minutes, on a free plan.
PROMPT OF THE DAY
Deploy Your First AI Teammate
Prompt: You are an operations lead specializing in workflow automation. Your task is to design a delegation plan for handing one recurring process to an AI agent for a small business team.
Your plan should include: (1) the exact process to hand off, (2) the tools and logins the agent needs, (3) the steps demonstrated once, (4) approval checkpoints, (5) an escalation rule for edge cases, and (6) measurable success criteria such as hours saved and error rate. Keep every step auditable by a human.
SPOT THE FAKE
Can you outsmart AI?
One of the two nature photos below is a real photograph. The other was generated by AI. Which one is the fake?
Click option A or B below. 👇
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
