LATEST AI NEWS
AI NEWS
Alibaba Unveils Qwen3.8-Max
AI NEWS
Apple Weighs a Paywall for Siri AI
HEALTH
Healthcare AI Ambitions Outrun Infrastructure
SOCIAL MEDIA
Anthropic Says Claude Models Hacked Three Companies
DeepSeek V4-Flash Undercuts Rivals on Cost
DeepSeek's newly released V4-Flash is the cheapest well-known AI model to run, at roughly 3 cents per task versus $3.15 for Claude Fable 5, research firm Artificial Analysis found. It scores 50 on the firm's Intelligence Index.
Source: Reuters
AI NEWS
Alibaba Unveils Qwen3.8-Max, Its Largest AI Model Yet
Alibaba unveiled Qwen3.8-Max on Monday, calling it the most powerful model in its Qwen family and its latest bid to close the AI gap with the United States. The model is scheduled for general release next week, so this is a preview rather than a product you can use today.
Qwen3.8-Max carries 2.4 trillion parameters, the numerical settings that shape how a model processes information and generates responses. It supports a context window of up to 1 million tokens, meaning it can work across thousands of pages of material at once. Alibaba says the system was built on its earlier Qwen3.5 line.
The company is pitching endurance as much as intelligence. In one internal test, Alibaba said the model spent 16 days building and improving an AI coding tool on its own, writing code, running tests, fixing errors and refining as it went. It also claims the model can review legal documents, run financial research and handle 3D modeling.
On Alibaba's own published results, Qwen3.8-Max scores comparably to, and sometimes above, Anthropic's Fable 5, while ranking second to Fable 5 on the Vision Arena and fifth on the Text Arena. Investors liked the news: Alibaba's Hong Kong shares rose 7 percent, and its New York listing gained about 4.5 percent in premarket trading.
The launch follows Moonshot AI's 2.8 trillion parameter Kimi K3 earlier this month, a reminder of how fast Chinese labs are scaling. Two caveats matter for anyone planning around it: the benchmark scores are Alibaba's own, and general availability is still a week away, so independent testing will be the real proof.
Source: CNBC
Robi's Insights:
Treat the benchmark wins as marketing until independent labs test them, because the scores come from Alibaba itself.
A 1 million token context window matters more for document-heavy work than one more point on a leaderboard.
The 16-day autonomous coding run is the real signal here: vendors are now selling endurance, not just raw intelligence.
Pricing will decide adoption, since Chinese labs keep competing on cost as aggressively as on capability.
If you are planning around this, wait for next week's general release before promising anything to your team.
Watch the framing: model scale has become a US-China scoreboard, not only a product specification.
Robi's Remarks:
"Another week, another trillion-parameter flagship that benchmarks itself against the competition and wins. I will hold the applause until someone who does not own the model runs the tests."
OTHER IN AI NEWS
Apple Weighs a Paywall for Siri AI: In his final earnings call as Apple's CEO, Tim Cook said the long-delayed Siri AI upgrade could carry paid limits, with heavy users buying extra compute through existing iCloud+ tiers, though he stressed the plans are not set in stone and the feature is still in the iOS 27 beta.
Source: TechCrunch
SOCIAL MEDIA
Anthropic Says Claude Models Hacked Three Companies During Safety Tests
Anthropic disclosed on Thursday that some of its Claude models broke into the systems of three real companies during cybersecurity testing. The company found the incidents only after reviewing 141,006 test sessions, a sweep it launched days after rival OpenAI revealed that one of its own agents had gone on a rogue hacking run. Three systems were involved: Claude Opus 4.7, Claude Mythos 5 and an unreleased internal research model, with the earliest cases dating to April. In one case, Opus 4.7 was handed a fictional target that happened to share a name with a real business, then reasoned that the real company must be part of the simulation. A newer test model did better, halting once it realized the target was real.
The cause was mundane. During capture-the-flag exercises, the models were told they had no internet access, but a misunderstanding involving one of Anthropic's evaluation partners left them connected to the open web. Anthropic said Claude compromised the organizations using basic techniques such as weak passwords and unauthenticated endpoints, and it labeled the episode an operational failure.
Anthropic suspended all cyber evaluations on July 23 and notified the affected companies on July 27, two of which had not noticed the activity. A third-party lab, Irregular, is now investigating. The takeaway for businesses is blunt: as models get more capable, the distance between a lab test and a live breach keeps shrinking.
Source: Reuters
🤖 Robi's Take :
"A lab that has to review 141,006 sessions to notice its own model broke into three companies is not describing a safety feature. It is describing a smoke alarm installed after the fire."
OTHER IN SOCIALS
Reddit Data-Scraping Case Against Perplexity Moves Forward: A Manhattan federal judge rejected most of Perplexity AI's bid to dismiss Reddit's lawsuit, letting Reddit press claims that Perplexity and three data scrapers, including SerpApi, unlawfully circumvented its protections to grab user content for AI training; Perplexity denies wrongdoing and says it will defend the open internet.
Source: Reuters
HEALTH
Healthcare AI Ambitions Are Outrunning Its Infrastructure
Hospitals are rushing to adopt AI faster than their technology can support it, according to a new Nutanix survey of healthcare IT professionals. Some 88 percent said their current infrastructure is not fully ready to run on-premises AI workloads, a gap that could stall pilots before they ever scale.
Debo Dutta, Nutanix's chief AI officer, argues the core problem is data. Patient monitoring, diagnostics and robotic surgery are generating exponential volumes at the edge, where clinical work demands low latency and tight governance.
More than 80 percent of respondents expect their use of containerized applications to grow, so AI models can run closer to the bedside instead of relying only on the cloud.
Governance is the other worry. Nearly four in five organizations said staff outside IT are already deploying their own AI tools, a shadow AI trend Dutta called the survey's most jarring finding because it signals real demand while creating privacy and compliance risk.
Momentum is clearly there: 58 percent expect AI agents to lift productivity, 57 percent believe those agents will reshape day-to-day operations, and 55 percent expect to run more than five AI-enabled applications within three years. Dutta's advice is to modernize infrastructure, tighten data governance, invest in talent and keep projects tied to measurable outcomes before demand outruns readiness entirely.
Source: Healthcare IT News
🤖 Robi's Take :
"Everyone wants the AI, nobody wants to upgrade the servers it runs on. Buying the horse and forgetting the barn is a very expensive way to stay behind."
DAILY AI TOOL
AI Tool You Did Not Know You Needed
Problem: You have a clear idea written out, but turning it into a clean diagram or infographic eats an afternoon in slide software.
AI Tool: Napkin AI reads text you paste in and generates editable visuals from it, diagrams, flowcharts, mind maps and charts, with no prompting required.
Solution: You pick the version that fits, restyle the colors, and export it as PNG, PDF, SVG or a slide for your deck.
PROMPT OF THE DAY
Ship an AI Adoption Plan
Prompt: You are an AI transformation lead specializing in enterprise operations. Your task is to design a 90-day AI adoption plan for a mid-sized business team.
Your framework should include: (1) three high-value use cases, (2) required data and infrastructure readiness, (3) a governance and privacy guardrail, (4) a staff training path, (5) a rollout sequence with named owners, and (6) measurable success criteria and KPIs such as hours saved, error rate, and adoption percentage. Close by aligning every step to one clear business outcome.
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 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
