The idea of using AI to help with computer programming has become a contentious issue. On the one hand, coding agents can make horrific mistakes that require a lot of inefficient human oversight to fix, leading many developers to lose trust in the concept altogether. On the other hand, some coders insist that AI coding agents can be powerful tools and that frontier models are quickly getting better at coding in ways that overcome some of the common problems of the past.

To see how effective these modern AI coding tools are becoming, we decided to test four major models with a simple task: re-creating the classic Windows game Minesweeper. Since it’s relatively easy for pattern-matching systems like LLMs to play off of existing code to re-create famous games, we added in one novelty curveball as well.

Our straightforward prompt:

Read full article

Comments

© Aurich Lawson | Getty Images

Recently, Forrester, a globally renowned independent research and advisory firm, released the report “Navigate The AI Agent Ecosystem In China, Forrester Research, October 2025[1].” NSFOCUS was successfully included in this report. In the report, Forrester identified four key technological trends: With the rapid advancement of Artificial Intelligence, AI Agent technology is deepening its application within […]

The post Unlocking New Possibilities for Security Operations: NSFOCUS’s AI Agent Capabilities Recognized by Authoritative Institution appeared first on NSFOCUS, Inc., a global network and cyber security leader, protects enterprises and carriers from advanced cyber attacks..

The post Unlocking New Possibilities for Security Operations: NSFOCUS’s AI Agent Capabilities Recognized by Authoritative Institution appeared first on Security Boulevard.

Google Labs’ new CC experiment uses Gemini to send a daily “Your Day Ahead” email that summarizes your schedule and tasks, and you can reply to it anytime for help.

The post Love the Now Brief on Galaxy phones? Google just built something better appeared first on Digital Trends.

An AI Agent whale has just turned one of this year’s loudest narratives into an expensive lesson, unloading a basket of agent tokens bought for $31.12M and getting back only $2.57M, according to on-chain data posted from Ember.

The wallet built its positions at the start of the year during the peak of the AI Agent narrative, when capital chased anything linked to autonomous trading bots and AI-powered execution.

With liquidity now thin and sentiment cooler, the same bets have been closed at a loss of about $28.54 million, or roughly 92%.

Token by token, the damage is stark. The whale lost about 91%, or $15.89M, on AIXBT and 92%, or $9.87M, on FAI.

太惨烈了~
一个巨鲸在年初的 AI Agent 热潮中用 $3112 万资金购买了多个 AI Agent 代币，如今随着浪潮退下，他在今天上午清仓割掉了这些 AI Agent 代币。他 $3112 万资金只卖回成 $257 万，亏损高达 $2854 万 (-92%)。$AIXBT ：亏损了 91% ($1589 万)；$FAI ：亏损了 92% ($987 万)；$NFTXBT… pic.twitter.com/P0KTfUdqXn

— 余烬 (@EmberCN) December 16, 2025

Aggressive Sell-Off Hammers AI Agent Tokens With Losses Across the Board

Positions in NFTXBT and POLY, both from the Virtuals ecosystem, were nearly wiped out with losses of 99%, equal to around $690,000 and $780,000 respectively.

The book did not fare much better elsewhere. BOTTO, an AI-driven art and curation project, produced an 84% hit of about $930,000. MAICRO, another Virtuals-linked agent token, cost the wallet roughly $380,000, a 90% drawdown versus its entry.

Because order books in these names are now shallow, the forced exit hit prices in real time.

Ember’s breakdown shows AIXBT falling about 10% during the selling, FAI dropping 8% and NFTXBT sliding 29%. BOTTO sank 32%, MAICRO tumbled 48% and POLY declined 26% as the whale worked its way through positions.

Arkham Reveals Large-Scale Sell-Off as AI Agent Hype Fades

Screenshots from Arkham’s explorer point to a sequence of transfers between the whale address and liquidity pools, with tens of millions of tokens in each project moving in quick succession. The flows suggest a deliberate decision to capitulate rather than a slow rebalance, locking in losses instead of waiting for a fresh burst of AI Agent speculation.

For market participants who rode the same theme, the episode is a reminder of how narrative-driven sectors can behave once attention shifts elsewhere. Many AI Agent tokens launched into the tail end of the broader AI mania and never built the depth or organic usage that support large tickets on the way out.

The liquidation also shows the limits of whale size in illiquid corners of crypto. Size that helps drive performance during the initial run-up can turn into a liability when liquidity dries up, since every attempt to exit pushes prices lower and erodes recovery value.

For traders still navigating the agents meta, the whale’s exit cuts both ways. It is a sharp reminder that late-stage narratives can punish even deep wallets, yet some may view the flush as clearing stale supply from thin markets.

The post Whale Unwinds AI Agent Positions at 92% Loss After Market Slump appeared first on Cryptonews.

Port offers a proprietary alternative to Spotify's popular devtool catalog Backstage, but with a valuable addition — it can be used to manage AI agents, too.

Kaspersky researchers uncovered Operation ForumTroll, an attack campaign utilising the new 'Dante' spyware developed by Memento Labs, the rebranded Hacking Team. The attacks used a Chrome zero-day vulnerability (CVE-2025-2783) and COM hijacking for persistence, confirming the continued deployment of advanced surveillance tools by the controversial Italian firm.