Senior AI Writer / Contributor
AI reporting, AI explainers
Google's September 2026 Pixel Drop, released September 15, adds a real-time "Likely scam" warning directly inside Gboard, Google's keyboard app, that appears above the keyboard when you start typing a reply to a suspicious message.
In a late-night X post on September 14, OpenAI CEO Sam Altman outlined two specific scenarios he says AI progress must avoid: humanity losing control of AI systems outright, and AI enabling an unprecedented concentration of power in a single person, company, or country.
Anthropic CEO Dario Amodei published an essay on Saturday, September 12, arguing that AI companies must slow the pace at which they improve model capabilities, citing the risk of losing control of increasingly self-improving systems.
Anthropic's September 2026 threat intelligence report, covering misuse activity between December 2025 and August 2026, identified an operation labeled GTG-54004 in which a single operator used Claude to mass-produce social media posts designed to look like spontaneous Kenyan public opinion ahead of the 2027 general election.
Nubia, ZTE's phone brand, will launch the NaviX Ultra in China on September 16, 2026, built jointly with ByteDance around the Doubao Phone Assistant, an AI agent designed to read the screen and complete multi-step tasks across apps from a single spoken request.
Researchers at UC Berkeley and Stanford published a framework called BeyondMimic in Science Robotics on August 26, 2026.
Google released a native Gemini desktop app for Windows 10 and Windows 11 on September 10, 2026.
Researchers at the Singapore University of Technology and Design, led by Assistant Professor Zhao Na, published HieraScaffold, an AI framework that generates large-scale, realistic 4D LiDAR environments, meaning 3D spatial data that also changes over time, for training autonomous vehicles.
Jacob Coxon, a 27 year old pretraining researcher who worked at both OpenAI and Anthropic over three years, resigned from Anthropic on September 9, 2026, and posted a seven part message on X warning that frontier AI labs are racing toward self improving superintelligence without adequate safeguards.
Apple Watch Series 12 and Watch Ultra 4 run on a new S11 chip and introduce Audio Intelligence, a feature that includes Siri Recap, which can summarize things like meetings based on audio notes you choose to capture.
Qualcomm's Snapdragon 8 Elite Gen 5 and MediaTek's Dimensity 9500 have both shipped with agentic AI built into their NPUs. Here's what's real and what's still rumor.
Netskope says GitHub and Microsoft OneDrive each affected 10% of European organisations through cloud malware distribution. Trusted SaaS is becoming attacker infrastructure.
Qualcomm and MediaTek are building hardware and software for AI agents that can reason and act locally, reducing cloud dependence for some tasks.
The next fraud problem is not simply better phishing copy. Autonomous agents could make social engineering cheaper to run at enormous scale, forcing financial institutions to automate defence too.
Anthropic is moving AI provenance into the output itself, but a detectable Claude mark can show model involvement without proving truth, plagiarism or human authorship.
South Africa's HyperDev, an AI coding platform founded by former OpenAI and Google engineers, raised over 1 million dollars in pre-seed funding as it approaches 100,000 users.
Kenya's MSME ministry is pushing AI adoption across millions of small firms as the Artificial Intelligence Bill 2026 moves through the Senate. The plan, the law, and the catch.
Older DDR4 platforms are back in the conversation because AI demand is squeezing memory supply and changing the real cost of a PC build.
Analyst estimates say Apple's first foldable iPhone could be scarce at launch. That may be a manufacturing constraint, but the sold-out effect still sells the product.
A token is a chunk of text an AI reads and writes, and a context window is how much of that text it can hold in mind at once.
An AI hallucination is when a chatbot states something false as if it were true, fluently and with total confidence.
An AI agent is software that does not just answer you, it takes actions to reach a goal you set.
For a lot of people, the free tiers are already enough, but daily users may benefit from a cheap mid-tier.
For two or three years, using powerful AI has felt almost free, but in 2026 that is changing.
