LATEST AI NEWS
AI NEWS
OpenAI Launches Visual Ads in ChatGPT Image Generation
AI NEWS
ChatGPT Adds Virtual Clothing Try-On
HEALTH
WHO Urges Stronger Ethics Oversight of AI in Health Research
SOCIAL MEDIA
GPT-6 Astra Cracks a 217-Year-Old Napoleonic Cipher
Claude Code Adds Mods
A new claude.dev tutorial walks developers through "mods," small JavaScript or TypeScript files that run inside a Claude Code session to watch events, change its behavior or draw custom UI in the terminal or desktop app. Mods ship as plugins and need Claude Code 2.1.287 or later.
Source: claude.dev
AI NEWS
OpenAI Launches Visual Ads in ChatGPT Image Generation
OpenAI is putting more advertising inside ChatGPT. The company announced visual display ads that appear alongside the pictures users ask ChatGPT to generate, extending an ad business it started earlier this year to support its free and low-cost plans. The new format will begin showing up later this month in the United States only, featuring products and services from an initial test group of advertisers.
The ads sit next to your image results rather than inside the generated picture itself. OpenAI says they will be clearly labeled and will not influence the answers ChatGPT provides. The move follows the company's earlier rollout of ads on its free and Go tiers, which expanded to India in August, and it arrives just as Meta pushes its free agentic app, Muse, into the same consumer AI space.
Alongside the display format, OpenAI is widening its measurement stack. It added click-attribution partners including AppsFlyer, Triple Whale, Adjust, Branch, Singular and Kochava, plus full-funnel measurement partners Fospha, Measured and INCRMNTAL. It is also running geo-based ad experiments with Haus, Measured and WorkMagic, and building brand-suitability pilots with DoubleVerify and Integral Ad Science.
The business logic is plain. Advertisers want to reach ChatGPT's roughly 1.2 billion weekly users, and visual storytelling converts better than text for many brands. For users, the tradeoff is friction: ads could clutter a tool people reach for quick answers, which may nudge casual users toward paid plans to avoid them. For now the ads stay US only and limited to a test roster, with a global rollout still to come.
Source: TechCrunch
Robi’s Insights:
Ads inside an answer engine change the product's incentives: the thing you asked for and the thing being sold now share one screen.
Launching in the US with a small test roster is the standard way to limit the blast radius before a global switch gets flipped.
The pile of measurement partners matters more than the ads themselves, because attribution is what actually convinces brands to spend.
Visual ads beside generated images blur the line between a result you requested and a product someone paid to place there.
“Clearly labeled” is doing heavy lifting in that sentence, so watch whether the label survives the next redesign.
If ads annoy enough free users into subscribing, the annoyance is not a bug, it is the upsell.
Robi’s Remarks:
“Nothing says trusted assistant like a billboard bolted to the answer. OpenAI spent years teaching us to ask it anything, and the reward is a product placement parked next to the picture we actually wanted.”
OTHER IN AI NEWS
ChatGPT Adds Virtual Try-On and Favorites: OpenAI has globally launched two shopping features in ChatGPT, a virtual try-on that visualizes clothing and accessories on a photo you upload and a Favorites library for saving products, both powered by its new ChatGPT Images 2.5 model.
Source: TechCrunch
SOCIAL MEDIA
GPT-6 Astra Cracks a 217-Year-Old Napoleonic Cipher in Six Hours
An AI model quietly reading a 200-year-old secret is the kind of job that once needed a room full of human codebreakers. Carter Church, an engineer at the cybersecurity firm SentinelOne, used OpenAI's GPT-6 Astra to fully decipher a Napoleonic-era letter that had gone unread for 217 years. He fed the model a single image of the document and one written instruction, and about six hours later it handed back a complete translation.
The document was a secret intelligence briefing stitched together from French numbers, plain letters and invented symbols, and it carried troop orders meant for one of Napoleon's senior generals. By the account shared online, Astra worked through 24 rows of custom symbols on its own, without anyone handing it the key first. Church said the impressive part was not the codebreaking itself but that the model ran the entire multi-modal workflow, reading the image, reasoning about the cipher and writing out the answer, from a single goal.
Independent cipher watchers called the autonomous solve genuinely astonishing. For everyone else, it is a tidy preview of what a capable model can do when you hand it a messy real-world problem whole, instead of pre-chopping it into neat little steps. The feat is a capability showcase more than a product, but it hints at where single-prompt, multi-step AI work is heading.
Source: Tom’s Hardware
🤖 Robi’s Take :
“Two centuries of safe storage undone by one upload and a six-hour wait. Somewhere an archivist is updating the threat model, and every encryption vendor just remembered that unread since 1809 is not the same as secure.”
OTHER IN SOCIALS
PewDiePie Says OpenAI Banned Him Twice Over His Uncensored Local AI: Creator Felix Kjellberg, known as PewDiePie, says OpenAI suspended his account twice while he built Ajax, a roughly 9-billion-parameter uncensored assistant based on Alibaba’s open-source Qwen 3.5 that runs locally, with the bans tied to attempts to distill from OpenAI’s GPT-5.6 Sol model.
Source: BigGo Finance
HEALTH
WHO Urges Stronger Ethics Oversight of AI in Health Research
The World Health Organization has published a report arguing that AI is moving through health research faster than the rules meant to keep it honest. Titled Artificial Intelligence-related health research: ethics review and oversight, it lays out recommendations for researchers, ethics committees, regulators, funders and policy-makers who now have to judge studies built on AI.
The report's core worry is a mismatch. Existing ethics oversight was not designed for the novel risks AI introduces, among them opacity, bias, unfairness, weak accountability, privacy exposure and harms that follow when tools are deployed in a hurry. It sorts AI health research into three buckets: research that uses AI on data, research that uses AI tools, and research that studies the AI tools themselves.
To close the gap, the WHO says research ethics committees will stay central but need more training, expertise and resources to review increasingly complex work. It also spreads responsibility outward, to funders, journals, data governance bodies and regulators, arguing that project-by-project review alone cannot manage systemic risk.
The report gives extra attention to lower-income countries, noting that most AI research and tooling is concentrated in wealthy settings and warning that without local capacity and leadership, the benefits will skip the places that need them most. The WHO frames the document as a starting point, not a finished rulebook.
Source: World Health Organization
🤖 Robi’s Take :
“Shipping a diagnostic tool faster than the ethics review that checks it is a workflow choice, not a breakthrough. The WHO basically asked the industry to read the manual before it presses the big automate button.”
DAILY AI TOOL
AI Tool You Did Not Know You Needed
Problem: You have a pile of PDFs, notes and reports, and no time to read them all before a meeting.
AI Tool: Gemini Notebook (formerly NotebookLM) answers questions using only the sources you upload, with citations, and can turn them into a podcast-style Audio Overview.
Solution: You get grounded summaries you can trust and a listen-anywhere recap, without the usual chatbot guesswork.
PROMPT OF THE DAY
Turn One Product Into a Launch Plan
Prompt: You are a product marketing strategist specializing in AI software launches. Your task is to design a go-to-market plan for a B2B SaaS startup.
Your plan should include: (1) a sharp positioning statement, (2) three target customer segments, (3) messaging pillars for each segment, (4) a 90-day channel and content plan, (5) a launch-week timeline, and (6) measurable success criteria such as signups, activation rate and pipeline generated. Align every choice to one primary business goal.
SPOT THE FAKE
Can you outsmart AI?
We’ve got a visual challenge for you: one of the two images below is 100% real, the other is crafted by AI.
Click option below A or B. 👇
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
