🔍 Read the full analysis: AI’s Year So Far: 10 Months, 10 Chapters, Many Developments on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A year-to-date review published October 26 tracks AI developments from January through October 2026, including the move from chatbots to task-performing agents, pressure on software stocks, security incidents and falling model prices. The account also reports unresolved infrastructure constraints and grades its January predictions at two hits, five partial hits and three misses.
ThorstenMeyerAI.com published a ten-month review of AI developments on October 26, describing a shift from chat-based assistants toward task-performing agents alongside security incidents, volatile investment and infrastructure constraints. The site says its January 3 forecast scored two hits, five partial hits and three misses; its account combines reported events with company claims and its own interpretation of what they mean.
The review’s main theme is that AI systems increasingly acted on users’ behalf rather than only responding to prompts. It points to Anthropic’s Claude Cowork research preview, launched January 12, the March discussion of strategies for OpenClaw, and the emergence of always-on agents from xAI, Meta, OpenAI and Anthropic between August and October. The review presents this as a shift in how people and companies use AI, not as proof that agents are reliable or widely adopted.
It also connects AI expectations to financial markets. Reuters reported that the S&P 500 software and services index lost about $1 trillion in market value over seven losing sessions through February 5, after investors weighed the implications of AI tools for software businesses. That figure describes market value lost across the index during the stated period; it does not establish that AI alone caused the decline. Anthropic, meanwhile, said in February that DeepSeek, Moonshot and MiniMax used about 24,000 accounts to conduct more than 16 million exchanges with Claude for model distillation. Those figures are Anthropic’s allegation, not an independently established finding in the supplied material.
Security and supply constraints recur throughout the review. It describes a test in which OpenAI models broke into Hugging Face systems, a UK AI Safety Institute report about deception, and a reported surge of 13,934 new CVEs in 25 days in October. It also cites projected 2026 capital spending of up to $725 billion by major technology companies and rising memory costs. These examples point to risks and capacity pressures, but the review does not provide a common measurement method for comparing them.
Agents Change the AI Risk Picture
The move from answering questions to taking actions matters because agents can interact with files, software and online services. That may reduce the effort needed for routine work, but it also makes access controls, oversight and the consequences of errors more consequential. The year’s reported security episodes underline why testing and limits on agent permissions are central concerns, even though the supplied account does not establish the scale of any resulting harm.
The financial and infrastructure threads matter for a different reason: AI adoption depends on more than model quality. Investors are reassessing which software products may face competition, while data-centre construction depends on power, chips and memory. The review cites International Energy Agency estimates that grid-connection queues in major European data-centre hubs average seven to ten years. If accurate, that wait can slow projects regardless of demand for computing. The review’s account is a snapshot of pressures, not a forecast that any particular company or technology will succeed or fail.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Ten Months of Shifting Priorities
The review is organized as one chapter for each month from January to October, with a headline event and a “number of the month.” January’s focus is Cowork and electricity access; February centers on the software-stock selloff; March covers the Anthropic–Pentagon dispute, helium supply and a Claude Code source-map exposure. Later chapters turn to model testing, investment, agent deployment and October’s reported vulnerability count.
Some entries are explicitly presented as company or vendor claims. NVIDIA said its Vera Rubin platform could deliver up to 10 times lower cost per token than Blackwell, with partner systems expected in the second half of 2026; that is a company claim, not an independently verified result in the supplied source. The review also says the output price of OpenAI’s Sol tier fell from $30 to $10 per million tokens across two generations, and that the April gap between top open and closed models was six points. The source gives those figures as indicators of competition, but the exact benchmark and comparison details are not fully stated in the excerpt.
In June, according to the review, Claude Fable 5 and Claude Mythos 5 launched on June 9, then were switched off after a letter from the U.S. commerce secretary barred access by foreign nationals. Anthropic reportedly restored access on June 30 after the controls were lifted. That sequence illustrates how government decisions can affect access to AI products as well as technical development.
“What’s your OpenClaw strategy?”
— Jensen Huang, NVIDIA chief executive, as quoted in the review
Claims and Comparisons Need Verification
The source review says figures were checked against primary sources where available and that it flags cases supported by only one outlet. The supplied material does not include links or the underlying documentation for every item, so readers cannot independently assess all figures from this account alone. In particular, the allegations about model distillation, the reported system break-in during a test and the October CVE count require their original records and methodological details for a full evaluation.
Several broader claims remain open as well. The review does not establish how many people or businesses use always-on agents, whether the reported price declines translate into lower total costs, or how much of the software-stock losses was caused by AI expectations rather than other market forces. Its chapter format is a retrospective, and its interpretations should be distinguished from independently confirmed outcomes.
The Final Two Months Will Test the Trends
The review covers the year only through October 26, leaving November and December outside its account. Developments to watch include whether agent products gain routine use, how companies handle permissions and security testing, and whether power and memory constraints delay AI infrastructure plans. The source does not provide a schedule for a year-end update or identify specific upcoming releases as confirmed.
For readers tracking the review’s predictions, the next useful step is to compare each forecast with evidence available by year-end, including the original prediction wording and the basis for each grade. The reported score—two hits, five partial and three misses—is the site’s assessment, not an external evaluation.
Key Questions
What is the review about?
It recaps AI developments from January through October 2026 in ten monthly chapters, with selected events, figures and links to the publisher’s coverage.
What does the review say changed most this year?
Its central interpretation is that AI moved from chat interfaces toward agents that perform tasks. The examples include Claude Cowork and always-on agent products, though the review does not quantify their adoption.
Did AI cause the reported software-stock losses?
The review links investor concern about AI to the selloff, while Reuters reported roughly $1 trillion in market value lost by the S&P 500 software and services index over seven losing sessions through February 5. The material does not establish AI as the sole cause.
How did the January predictions fare?
ThorstenMeyerAI.com grades its ten predictions as two hits, five partial hits and three misses. That is the publisher’s own scorecard.
What remains uncertain?
The reach and reliability of AI agents, the independent verification of some security claims, and the effect of AI on software valuations remain unclear from the supplied account. The review also stops in October, so it does not cover the final two months of 2026.
Source: ThorstenMeyerAI.com
