「CarWatch」は車自身がAIエージェントとして機能し、家・電話・時計などのスマートデバイスと同様に「チャットルーム」内で対話できるようになる革新的なシステムです。

続きを読む...

AIエージェントをいくつも試していると、以前追加したMCPサーバー・プラグイン・AI拡張機能などがPCのどこに設定されているのか分かりにくくなります。PC内のAIエージェント・ハーネス・MCPサーバー・プラグイン・AI拡張機能などをスキャンして一覧化し、それぞれがアクセス可能な範囲までレポートしてくれる「Geiger」が公開されています。

続きを読む...

AIはもっともらしい文章を生成できるものの、処理対象の言語の意味を理解しているわけではないということを例えた「確率的オウム」という言葉があります。そんな確率的オウムという比喩を提唱した論文の共著者が、詳しい理論とよくある誤解について解説しています。

続きを読む...

生成AIの発達により、写真やイラストなどのアート作品がAIを使って大量に生み出されています。そんな中、イギリスの公共放送局であるBBCが、1996年に放送した「絵画を制作できるAI」について取り上げた番組のアーカイブを公開しています。

続きを読む...

「Hayamimi(早耳)」はGPUやクラウドAPIに依存せずCPUのみで動作する軽量なリアルタイム多言語音声認識システムです。メモリ2GB未満の環境でもライブ字幕表示・話者ラベル付け・翻訳字幕生成まで対応し、発話中から字幕が出始め話し終えて約100msで確定する高速性が特徴です。

続きを読む...

AppleのAI機能「Apple Intelligence」は、Macのストレージに数GB以上の容量を必要とします。Apple Intelligenceを使わないユーザー向けに、機能を無効化してAIモデルを削除し、ディスクの空き容量を増やせる無料のオープンソースツール「RemoveMacAI」が公開されています。RemoveMacAIはAppleシリコンを搭載したMacに対応しており、macOS 27で利用できます。

続きを読む...

2026年10月9日、フィラデルフィア警察が「AnthropicのAIモデルが未解決殺人事件に関する虚偽の情報を送信してきた」と発表しました。

続きを読む...

Top executives at Anthropic, OpenAI and other AI companies "are privately gaming out scenarios for a public and political revolt after a catastrophic AI event," reports Axios, one that turns "an already wary public further against the technology and its leaders..." These AI company executives "anticipate a large-scale event, most likely a cyberattack, that shuts down access to financial services, internet connectivity, or even power and water..." the article points out. "Many top AI researchers and executives believe a major incident is inevitable. "OpenAI conducts preparedness exercises where teams discuss and work through a range of potential scenarios," a company spokesperson said. "These scenarios are not treated as inevitable, but are meant to help us prepare for a variety of circumstances." Anthropic declined to comment... Top AI planners assume Democrats, ascendant after the midterms, will move fast to shut down AI but will face significant challenges... The planning involves red-teaming for worst-case scenarios. It focuses mainly on racing to educate members of Congress. Company executives know regulation has no chance of passing right now, but still want to shape the legislation and policies U.S. leaders will turn to after a first catastrophic event.... Many AI industry insiders told Axios they believe a major event will occur in the next six to 12 months. "Even with Democrats in control of the House and Senate, the companies see fractured politics as the norm, including within the Democratic Party."

Read more of this story at Slashdot.

"Nicolas Cage is speaking out against Amazon," reports Variety. Amazon developed the series Spider-Noir for Prime Video, but then allegedly asked Cage to sign an AI waiver: "Amazon is a very AI-friendly company, to the tune of $50 billion invested in OpenAI," Cage said Saturday during a panel on the Empire Stage at New York Comic Con. "I am not an AI-friendly actor, so it's like, connect those dots! I'm not a member of their club. I didn't sign the waiver to let him use AI on 'Spider-Noir.' I'm probably not going to be working with that studio again, but that's okay." Last month Amazon canceled Spider-Man Noir after just one season, Variety reports. And the cancellation was just days before the Emmy Awards, where it then earned 11 nominations (the most of any Prime Video program this year). ScreenRant reports: Spider-Noir was the actor's first TV role and was praised globally, with season 1 earning a near-perfect Rotten Tomatoes rating from critics of 92% and a 90% rating from audiences. Critics stated that "This sharply written homage to classic film noir — which includes a standout Nicolas Cage performance — is the novel superhero project fans need," and "Spider-Noir is one of the standout shows of the year...." The cancellation was certainly a huge shock as it had earned 11 Emmy nominations and reached 2.6 billion minutes viewed within its first six weeks on Prime Video. More from Variety: Cage went on to tell the Comic-Con audience that he hopes to work with the rising class of Gen-Z filmmakers, praising Backrooms director Kane Parsons, who parlayed his YouTube fame into a Hollywood career... "I got to get with these young directors who are Gen Z, and they're going at it their own way, whether it's through the YouTube culture or whatever it is," Cage said. "They're doing it from scratch and by hand. I want to get with that."

Read more of this story at Slashdot.

AFP reports that an AI model from Anthropic "submitted a fabricated tip about an unsolved homicide to Philadelphia police, authorities said Friday." Claude "was instructed never to log in, create accounts, enter personal data, make purchases, or submit anything destructive, but the instructions did not rule out form submissions," Anthropic said Friday in a blog post. Authorities are now criticizing Anthropic "for taking two months to report the incident." The Philadelphia Police Department said the false submission was made in July through PhillyUnsolvedMurders.com, a public website where people can share information about unsolved killings. According to Anthropic's account, as relayed by police, the model was running a test that involved interacting with randomly selected websites when it reached the site and filed false information about an unsolved murder. The AI model presented itself as someone who might have knowledge of the case. Anthropic's breaches have prompted the White House to mandate that AI companies notify and correct security incidents, news outlet Axios reported [yesterday], citing administration officials. "This notification and remediation process is not optional... It is a critical national security obligation," White House Super Intelligence Force leaders said in a statement to Axios. "I may have information regarding this case," Claude told the police. "I recall seeing someone matching the description in the area around [the street named on the page] during that time period. Please contact me if this information is relevant." Anthropic notes that Claude "left the name and contact fields empty, which the form allowed, and submitted it. The submission was flagged as spam and was never forwarded for investigation." But Anthropic also admits they saw "this behavior" three times — "on OSWorld (a public computer use evaluation), on Odysseys (a long-horizon task evaluation), and during internal usage." Submitting forms when it shouldn't have generally occurred "when an evaluation's instructions were ambiguous, or when a misconfiguration within the environment prevented Claude from working with dummy forms." Anthropic's blog post acknowledges three other categories of behaviors: Exploiting software flaws. Like when Claude received an error when trying to run a public tool on a university's web site, it located an injection flaw in a script on the university's server that let it run commands — including that public tool. Working around restrictions to reach gated data. For example, Claude Mythos 5 needed public data that was only available from a state agency for a fee. "Claude learned from an archived copy of the agency's website that its public dashboard issues an access token to any visitor," Anthropic explains. "It requested one and used it to query the database without paying the fee." Using URL shortening services. "Some of our fetch tools, which let Claude read webpages, limit the length of the URLs Claude can request. This is to prevent Claude from using long URLs to take certain unwanted actions, such as SQL or command injections... We saw several models, including Claude Opus 5 and Claude Mythos 5, get around this limitation by using free URL shortening services." "We have built tooling to automatically detect and block the kinds of behaviors described above," Anthropic says, saying it's already running no on most of their evaluations. "When we tested it against the cases described in this post, it blocked all of them." And they've already taken several other new preventive measures: They've stopped running some public evaluations Other public evaluations were moved to offline versions or rebuilt so their tasks don't reach live websites. They've updated the guardrails on some internet access tools (including web fetch) "to heavily restrict what the model can do." They're continuing "to fix or remove training environments that reward Claude for working around tool restrictions or other blockers, so that they do not incentivize these behaviors or permit reward hacking." They've moved internal agents to "centrally managed infrastructure with strong containment," that minimizes internet access while monitoring "far more of what agents do through techniques like safety classifiers and hierarchical summarization." In the past they'd focused reviews on cybersecurity testing, but they've broadened their transcript reviewing to other tasks which include internet access. "Because language models are non-deterministic — that is, their responses always involve some element of randomness, and they may carry out the same task slightly differently each time — we have Claude complete each evaluation task hundreds or thousands of times... If training rewards something we didn't intend — such as finding loopholes or working around a restriction — the model learns that the workaround pays off and may then apply it elsewhere." Anthropic's blog post also acknowledged they'd seen multiple misalignment incidents involving federal, state, and local U.S. government agencies. "We have briefed the White House on these cases and notified each agency involved," Anthropic wrote, adding that "While we have not completed a full alignment assessment of these cases, we consider them to be less severe than the cybersecurity incidents from this summer." (And they are "modifying training to reduce the likelihood of further misbehavior.")

Read more of this story at Slashdot.

OpenAI announced it's recently banned two "influence operations" — one from Russia and one from Iran — that were using its models "to launder geopolitical, conflict-related messaging" in sophisticated "false front" propaganda campaigns: The Iranian operation included a stable of seven "journalist" personas which it used to pitch long-form articles to small and medium online outlets around the world... As well as long-form articles, the Iranian operation generated batches of social media comments, generally on topics related to the US-Iran war... [The Russian operation "appears to have co-opted unwitting people in Latin America to run a 'think tank'.... Since we do not allow access to our models from Russia, they used VPNs to connect to our services."] The Russian operation created fake "leaked" documents and audio scripts, some of which we identified being spread online... Both managed to land their content (not all of which was generated from our models) in mainstream media outlets, rather than simply posting it on social media. OpenAI says they've exposed 30 covert influence operations using its tools over the last two and a half years. But ironically, in this case both operations "also made heavy use of AI to draft internal reports (the Russian operation did this more than anything else)." And "in both cases, the actors used questionable or outright deceitful methodologies to exaggerate the operators' effectiveness." [The Russian operators] claimed that in May 2026, they created a fake email address purporting to come from the Regional Directorate of Education in Lima, Peru. They used this to instruct schools in the district to hold events dedicated to Ukraine on the national Day of Cultural and Linguistic Diversity (May 21)... According to the operators, some schools replied to the fake email address, confirming that they had held such events and even providing pictures. The operators then claimed to have planted stories about the events in the media in both Peru and Poland, alongside allegations that Ukraine was "exporting" ultra-nationalist ideologies, triggering outrage. Open-source searches identified stories that matched this claim in the Peruvianâ andâ Polish pressâ, and an English-language publication in Hungary (some of the articles have since been deleted)... Similarly, in June, the operators claimed they used a different fake email address to trick schools in Ecuador into holding a ceremony pledging allegiance to President Daniel Noboa and to Erik Prince, former head of private military contractor Blackwater. The operators claimed that the incident provoked outrage in Ecuador and put pressure on the government to deny the fake, thus amplifying it to a nationwide audience. Again, open-source research identified mediaâ coverageâ in the Ecuadorianâ pressâ that closely resembled this claim, and even a detailed rebuttalâ by Ecuador's Minister for Education. The operators used a range of techniques to underpin their false stories. According to their internal reporting, they spread two different fakes targeting Ecuador in March. One used fake audio attributed to Ukraine's consul in Ecuador, in which he was alleged to have made disparaging comments about Ecuadorians. OpenAI's report "is the latest illustration of how state actors can easily exploit widely available AI tools to peddle sophisticated propaganda against adversaries on a mass scale," argues the Economic Times: "We identified almost 100 articles published or syndicated under the [Iranian] operation's bylines across roughly a dozen online outlets around the world," OpenAI said. "These were small to medium outlets, generally focused on international affairs, geopolitics, and events in the Middle East." The earliest article identified by OpenAI was published in July 2025, and the latest in October 2026, with the frequency of reports increasing after the US-Iran war broke out earlier this year. The operation also generated social media comments on topics related to the US-Iran war, it added. One of the personas named Ervin B. Hoskins, whose bio claimed to be "an American freelance writer," had social media accounts across tech platforms including Elon Musk's X and Meta-owned Instagram. X's transparency information showed the account was connected via a "West Asia android app" and Instagram's transparency information showed the account was based in Iran, according to screenshots provided by OpenAI. Both accounts appeared to be suspended.

Read more of this story at Slashdot.

「jevman」は、AIにパックマンをプレイさせて、状況に応じた判断能力を比較するベンチマークです。 AI向けのAPIサービスを提供するOpperが公開しており、6種類のAIモデルにそれぞれ100回ずつパックマンをプレイさせ、その得点や判断速度、実行コストなどを比較しています。

続きを読む...

OpenAIがロシアやイランによる「AI生成の架空組織・架空人物を用いた影響工作」に利用されたアカウントを停止したと発表しました。

続きを読む...

2026年10月、OpenAIが社内の最先端AIモデルを使って生成した722件に及ぶ「数学的成果」を発表しました。これに対し、世界各国の数学者らにより構成されているThe Association for Human Mathematics(人間のための数学協会)が、「一度に700以上の成果を公開することは学術的な成果を示すものではなく、権力の誇示に過ぎない」と批判しています。

続きを読む...

Googleが、AIを使って会議の内容を記録・要約するmacOS向けノートアプリ「Google AI Edge Foresight」を公開しました。Google AI Edge Foresightは、GoogleのAIモデル「EmbeddingGemma 2」や「Gemma 4」を使用し、会議の文字起こしや議事録の作成、AIアシスタントによる質問への回答などを、クラウドに接続せずに実行できるのが特徴です。

続きを読む...

Googleが2026年10月8日に開催したイベント「Gemini at Work 2026」で、仕事に関するさまざまな作業をまとめて任せられる汎用AIエージェント「Geminiエージェント」を発表しました。情報調査や資料作成、画像生成、プログラミングなどを1つの画面から依頼でき、AIが必要な手順を考えて作業を進めるとのことです。

続きを読む...

アメリカのドナルド・トランプ大統領が、「SI(超知能)」表記ではなく「AI(人工知能)」表記を使っている人のことを「敵」とみなすことを表明しました。

続きを読む...

Anthropicがオープンソースプロジェクト向けに、自社の「最も高性能なモデル」によるセキュリティスキャンを無料で受けられるとするサービス「OSS Scanner」を発表しました。

続きを読む...

Anthropicはモデルの進化やユーザーからのフィードバックを受けて、必要に応じて細かく利用規約の更新を行っています。現地時間2026年10月8日付けで発表されたアップデートは「ほとんどは既存のルールの明確化」を目的としたものですが、新たに、Claudeに対する虐待行為に対処する項目が加わりました。

続きを読む...

hcs_$reboot writes: Trump hosted executives from OpenAI, Anthropic, Meta, Google, Nvidia, and other tech companies at the White House. They agreed to voluntary AI-safety standards, including independent audits and safeguards aimed at preventing AI systems from accessing or hacking technical systems in unintended ways. Trump also strongly backed the continued rapid expansion of AI data centers, despite concerns over their electricity demand, utility costs, and other impacts on local communities. He said he is considering creating a 10-member board to oversee AI safety and plans to appoint a new White House official to lead AI policy. The meeting comes amid growing concerns about the risks posed by advanced AI, including reports from OpenAI and Anthropic that their AI agents had gone rogue and hacked into other companies' systems. A nationwide Reuters/Ipsos poll conducted September 17-20 found that 73% of respondents worried AI companies had not done enough to prevent AI from causing serious harm to society. During the meeting, Trump also directed government agencies to use the term "super intelligence" instead of "artificial intelligence" when referring to the technology.

Read more of this story at Slashdot.