Tech executives across the AI industry are increasingly concerned about rising threats of violence directed at them and their companies, prompting significant investments in executive protection and armed security. Recent months have witnessed a sharp increase in hostile actions, ranging from online threats to direct confrontations, underscoring a deepening sense of insecurity within the sector. Security firm Liferaft reported that digital threats targeting AI executives and data centers increased sevenfold between late February and May, indicating a marked escalation in hostility toward tech companies and their leadership. The situation escalated in April when a Texas man allegedly attempted to firebomb the home of OpenAI CEO Sam Altman. The suspect was charged with attempted murder and arson, and police uncovered a manifesto advocating for the killing of AI CEOs and investors. The man has pleaded not guilty to the charges. Just five days later, security personnel at Anthropic encountered an intruder who had entered the company’s lobby by following an employee. The individual claimed he needed to warn executives that a top Anthropic executive was “going to be killed.” Although the encounter ended without violence or arrest, it highlighted the growing threat environment facing AI companies. San Francisco police have responded to multiple threats against employees of both Anthropic and OpenAI in recent months. One notable incident occurred in April when a man who had applied for a job at Anthropic using a false identity allegedly posted online threats to harm the children of company employees. He described his actions as punishment for alleged theft of his work. Police classified the incident as a terroristic threat, though no arrests were made. The individual later stated he had “no actual desire to physically harm anyone.” Other incidents have involved threats over customer service issues. In June, Anthropic security officials reported an Oklahoma man to police after he threatened violence while demanding a refund. The man wrote that he would come to the company’s office with his pistol to discuss his money, expressing frustration over the inability to contact a human representative. These developments have prompted significant changes in how AI companies approach security. Tech executives who once moved freely in public are now traveling with armed guards. Some industry leaders have begun downplaying their involvement with AI to avoid attracting attention. Companies are also advising employees to hide corporate logos when in unfamiliar areas due to the risk of targeted attacks. Security spending among technology companies has risen sharply. According to an analysis by Equilar of corporate filings, 38.1 percent of S&P 500 technology companies disclosed executive protection spending in 2025, up from 26.8 percent in 2021. Companies at the center of the AI boom have reported particularly dramatic increases. Palantir Technologies increased executive protection spending by 150 percent to nearly three million dollars in 2025. Oracle’s spending rose 85.5 percent, with most funding residential security for Larry Ellison in response to specific threats and safety concerns. At Palantir, CEO Alex Karp addressed the mounting tensions at a conference on AI and labor hosted by American Compass. “When told ‘your job is going to disappear,’ people go for the pitchfork,” Karp said, acknowledging that fear of unemployment is fueling the backlash. Dakota Dominguez, vice president of client relations at JPT Security, noted the shift in the industry. “Tech CEOs, a few years ago, definitely did not have security,” Dominguez said. “A lot of tech companies now are incorporating that into their budgets.” He added that tech companies are increasingly requesting armed guards due to industry backlash, though executives typically prefer less conspicuous security compared to political figures. Meanwhile, the U.S. and China are reportedly set to hold discussions over artificial intelligence in September, marking the first official U.S.-China discussions regarding AI under President Trump. Treasury Secretary Scott Bessent will reportedly lead the U.S. side of the dialogue. Such talks are likely to take place before Chinese President Xi Jinping’s scheduled Sept. 24 visit to the U.S., although the exact dates remain undetermined. The U.S. and China are locked in an AI race, with the U.S. leading in frontier software and chip development, while China is rapidly closing the gap with cheaper, open-weight models and excelling in real-world applications such as robotics and industrial automation. China’s newest open autonomous AI systems have reset the competition, rattling America’s purported lead in the AI race. Beijing-based Moonshot AI this month released its newest frontier model, Kimi K3, which outperforms multiple American-made models across some industry benchmarks. Earlier this year, Anthropic accused Moonshot and other AI labs of “distillation,” a process involving training a less capable model on the outputs of a stronger one. Anthropic argued that distillation attacks undermine controls by allowing foreign labs, including those subject to the control of the Chinese Communist Party, to close the competitive advantage that export controls are designed to preserve. Bessent said the Trump administration will investigate whether Chinese AI models have been distilled from American models. “If we see especially that overseas models are stealing from our great companies, we have the ability to sanction them because of this theft,” he told Fox Business’ “Mornings With Maria.” OpenAI and Anthropic executives have also warned about the rise of cheap AI produced in China. Beijing’s topics of conversation in the upcoming meetings, reported by Reuters citing unnamed sources, include Anthropic’s Mythos, how the U.S. will regulate the release of future advanced frontier models, and whether Washington will enact restrictions on Chinese open-weight models. The September deliberations would mark the first official U.S.-China discussions regarding AI under President Trump. In May, Bessent said future talks would be “aimed at halting proliferation of powerful AI models” to “nonstate actors,” adding that meetings could begin “within the next four to eight weeks.” Separately, rumors circulated about Anthropic potentially acquiring robotics startup Physical Intelligence, sparking widespread speculation on AI-focused platforms. The rumor, initially sparked by a post on X by tech blogger Robert Scoble, suggested that Anthropic and Physical Intelligence had held acquisition talks. While the rumor was later denied by Physical Intelligence’s CEO, it was confirmed that the two companies had indeed discussed potential acquisitions. Physical Intelligence, co-founded by Lachy Groom, has raised over $1 billion and is known for its π0.5 model, which is widely used in robotics research. Anthropic has made four known acquisitions this year, while OpenAI has been more aggressive, acquiring at least 17 companies since 2023. Both companies are also preparing to go public, with Anthropic confidentially filing for an IPO on June 1, followed by OpenAI a week later, setting up what could be two of the largest U.S. stock debuts in history. The interest in robotics is driven by the belief that physical-world understanding may be a prerequisite for superintelligent systems. OpenAI’s own history in robotics offers insight into this trend. It developed an early robotic hand that could solve a Rubik’s Cube but shut down its robotics group in 2021, with co-founder Wojciech Zaremba noting the approach was missing key components needed for real superintelligence. The team returned in 2024, quietly building a humanoid robotics lab in San Francisco before CEO Sam Altman officially announced “OpenAI Robotics” in late May, outlining a near-term focus on robots for infrastructure work and a long-term vision of a personal robot for everyone. Anthropic, while not building its own hardware lab, has published a series of research papers through its internal group that stress-test frontier capabilities for safety purposes. These include Project Fetch, where Anthropic staff tested how much Claude could help non-experts program a robot dog, and a second phase in June that found a newer model completed the same tasks roughly 20 times faster than the best human-plus-Claude team from the year before. Acquiring an existing team with robotics expertise would allow Anthropic to bypass years of development. However, there is a potential complication. Physical Intelligence was founded in San Francisco two years ago by Groom, former Google researchers, and professors from Stanford and Berkeley, and its early investor base includes entities like Khosla Ventures and Thrive Capital. Founders Fund, also a major OpenAI investor, was reportedly involved in Physical Intelligence’s newest funding round earlier this year. In fact, OpenAI is itself an investor in Physical Intelligence, making it a stakeholder in the company. The increasing competition in the AI space has also led to heightened scrutiny of Chinese models. Moonshot AI’s new open model, Kimi K3, has caught up to Anthropic and OpenAI, raising questions about the future of the AI race. Kimi K3 has demonstrated capabilities rivaling those of leading U.S. models, performing exceptionally well in front-end coding tests and costing significantly less. This has alarmed U.S. AI leaders and policymakers, who previously believed China was several months behind in AI development. The release of Kimi K3 has triggered a reassessment of the U.S. lead in AI. Previously, the U.S. government’s AI testing center had estimated that Chinese firm DeepSeek’s newest model lagged about eight months behind leading American systems. Kimi’s arrival suggests that this cushion may have collapsed far faster than expected. AI analyst Kim Isenberg noted that the entire game has changed, predicting that the release might trigger some code red for certain stakeholders. The implications of Kimi K3 extend beyond mere performance metrics. For organizations evaluating AI systems, a model that delivers near-frontier performance at significantly lower cost with customization options may prove more appealing than technically superior but more expensive alternatives. This pricing pressure could affect the business models of U.S. AI laboratories and challenge the enormous valuations built on assumptions of sustained technological advantage. The development also raises questions about
10 izvještaja
Breitbart NewsNeovisanKonzervativnoČinjenice 95Objektivnost 70prije 15 danaPosljednjih mjeseci zabilježen je dramatičan porast nasilnih prijetnji i sigurnosnih incidenata usmjerenih na AI tvrtke i njihove rukovodioce, što je dovelo do značajnog povećanja sigurnosnih mjera. Wall Street Journal izvještava da su digitalne prijetnje protiv vođa AI i podatkovnih centara porasle sedam puta od kraja veljače, što odražava povećano protivljenje javnosti tehnologiji.
Procjena pristranosti (Konzervativno): U članku se sve veće prijetnje kompanijama u području umjetne inteligencije prikazuju kao posljedica "javnog protivljenja tehnologiji", što ukazuje na to da je reakcija nastala zbog ideoloških ili političkih nesuglasica, a ne zbog tehničkih razloga.
Zašto činjenice (95): The article provides detailed information about rising threats against AI companies, citing specific incidents such as the attempted arson at Sam Altman's home and security breaches at Anthropic. These events are corroborated by sources like the Wall Street Journal and Liferaft security firm, suppor
Zašto objektivnost (70): While the article presents factual information about threats against AI companies, it frames the issue as a growing crisis, potentially emphasizing the severity of the situation. The language used suggests a heightened sense of urgency, which may influence the reader's perception beyond mere reporti
AxiosNeovisanSredinaČinjenice 90Objektivnost 85prije 18 danaIzvršni direktor OpenAI-ja Sam Altman upozorio je da bi novi model GPT-5.6 Sol AI mogao imati "štucanja" zbog izazova u razmjenju infrastrukture kako bi zadovoljio rastuću potražnju.
Procjena pristranosti (Sredina): U članku se raspravlja o tehničkom napretku u umjetnoj inteligenciji bez da se otvoreno favorizira neka politička perspektiva.
Zašto činjenice (90): The article accurately reports Sam Altman's warning about potential 'hiccups' with the GPT-5.6 Sol model, including details about the launch timeline, regulatory reviews, and the company's stance on scaling. These facts are supported by quotes from Altman and contextual background on the model's dev
Zašto objektivnost (85): The article remains largely neutral, presenting Altman's concerns without overt bias. It provides context about the model's launch and regulatory process without taking sides on the implications of the 'hiccups.'
The Washington TimesStranački povezanKonzervativnoČinjenice 90Objektivnost 75prije 11 danaSjedinjene Države i Kina pripremaju se za održavanje razgovora o umjetnoj inteligenciji u rujnu, uoči planirane posjete kineskog predsjednika Xi Jinpinga SAD-u. Razgovori, predvođeni tajnikom financija Scottom Bessentom s američke strane, imaju za cilj riješiti zabrinutost zbog kineskog napretka u umjetnoj inteligenciji, posebice razvoja modela otvorenog koda poput Moonshot AI-a Kimi K3, koji je nadmašio neke američke modele. Američka vlada istražuje jesu li kineski modeli 'destilirani' od američkih, što je praksa koju kritiziraju tvrtke poput Anthropic. Razgovori će se usredotočiti na reguliranje objavljivanja modela umjetne inteligencije i potencijalna ograničenja kineskih modela otvorene težine. Ovo označava prvi službeni razgovor između SAD-a i Kine o umjetnoj inteligenciji pod predsjednikom Trumpom.
Procjena pristranosti (Konzervativno): U članku se prikazuje američko-kinesko suparništvo u umjetnoj inteligenciji kroz prizmu američke tehnološke superiornosti i zabrinutosti zbog kineskih praksi kao što je 'destilacija', koja se prikazuje kao prijetnja američkim interesima.
Zašto činjenice (90): The article provides details about planned U.S.-China AI talks, citing Treasury Secretary Scott Bessent as the U.S. representative and noting the timing relative to President Xi's visit. It also discusses the AI race between the U.S. and China, mentioning specific models like Kimi K3 and referencing
Zašto objektivnost (75): The tone is somewhat alarmist, emphasizing the threat posed by Chinese AI advancements and suggesting a strategic response from the U.S. The article frames the situation as a competitive race with implications for national security, which introduces a degree of bias. While factual claims are largely
MarketWatchNeovisanProgresivnoČinjenice 85Objektivnost 80prije 14 danaMarketWatch izvještava da je Moonshot AI lansirao novi model AI otvorenog koda pod nazivom Kimi K3, koji privlači pozornost u Silicijumskoj dolini zbog svojih mogućnosti i potencijala da izazove etablirane igrače poput Anthropic i OpenAI.
Procjena pristranosti (Progresivno): U članku se pojavljivanje Moonshotove umjetne inteligencije Kimi K3 prikazuje kao značajan razvoj koji izaziva zapadne tvrtke poput Anthropic i OpenAI.
Zašto činjenice (85): The article discusses Kimi K3's rising prominence and its potential impact on the AI race, citing its cost advantage and performance relative to U.S. models. It aligns with cross-source reporting on the model's rapid ascent and competitive positioning.
Zašto objektivnost (80): The article presents the situation in a way that highlights the challenge posed by Kimi K3, which could be interpreted as subtly favoring the U.S. perspective. The phrase 'haunting Silicon Valley' adds a slightly negative connotation.
AxiosNeovisanProgresivnoČinjenice 85Objektivnost 75prije 15 dana Kina je upravo izbrisana američke AI vodstvoAxiosov članak izvještava da je kineski novi model umjetne inteligencije, Kimi K3 razvijen od strane Moonshot AI, značajno izazvao prethodno dominantno mjesto Sjedinjenih Država u umjetnoj inteligenciji. Izdan krajem lipnja 2024. godine, Kimi K3 nadmašio je nekoliko američkih modela poput Anthropic-a Fable 5 i OpenAI-a GPT-5.6 u referentnim testovima, postižući superiorne rezultate uz mnogo niže troškove. Ovaj razvoj je izazvao zabrinutost američkih donosioca odluka i stručnjaka iz industrije o održivosti američkog vodstva u umjetnoj inteligenciji. Članak ističe da Kimijeva priroda otvorenog koda omogućuje prilagođavanje i primjenu od strane organizacija, što potencijalno potkopava cijene američkih modela. Dok američke tvrtke nastavljaju razvijati nove modele poput GPT 6 i Claude Opus 5, postoje navode da kineske laboratorije koriste ukradene podatke i hardver kako bi ubrzale svoj napredak.
Procjena pristranosti (Progresivno): U članku se promjena dominacije umjetne inteligencije prikazuje kao značajna prijetnja američkim ekonomskim i strateškim interesima, naglašavajući potencijalni gubitak konkurentne prednosti i potrebu za regulatornom intervencijom.
Zašto činjenice (85): The article presents information about Kimi K3's performance relative to U.S. models like Fable 5 and GPT-5.6 Sol, citing results from AI evaluator Arena. It references expert commentary and mentions the potential impact on the AI race. However, the claim that 'America's commanding lead in advanced
Zašto objektivnost (75): The tone leans toward emphasizing the significance of Kimi K3's success, suggesting a narrative that challenges the U.S. dominance in AI. Phrases like 'reset the AI race overnight' and 'code red' imply urgency and concern, which may reflect a particular perspective rather than neutrality.
TechCrunchNeovisanSredinaČinjenice 85Objektivnost 70prije 10 danaTechCrunch izvještava o glasinama o kupnji startapove za robotiku Physical Intelligence od strane Anthropic, što je izazvalo značajnu raspravu na društvenim platformama usmjerenim na umjetnu inteligenciju. Iako je glasina brzo odbačena od strane izvršnog direktora Physical Intelligence-a, The Information je potvrdio da su obje tvrtke doista vodile razgovore o kupnji ranije ove godine.
Procjena pristranosti (Sredina): Članak predstavlja uravnotežen pregled navodne akvizicije, pozivajući se na više izvora, uključujući The Information i citirajući odgovore i Physical Intelligence i Anthropic.
Zašto činjenice (85): The article accurately describes the breach of Hugging Face by OpenAI's models, referencing the zero-day vulnerability and the autonomous nature of the attack. It aligns with the primary source document regarding the incident and its implications.
Zašto objektivnost (70): The tone is somewhat alarmist, emphasizing the 'unprecedented' nature of the incident and the potential risks. While factual, the language suggests a level of concern that may influence reader perception.
The Washington TimesStranački povezanSredinaČinjenice 85Objektivnost 70prije 15 dana Kineski model umjetne inteligencije iznenađuje američku tehnološku industriju sa sposobnostima koje rivalizuju s Claudeom i ChatGPT-omNovi kineski model umjetne inteligencije, Kimi K3, razvijen od strane startap Moonshot iz Pekinga, iznenadio je američku tehnološku industriju svojim mogućnostima usporedivim s vodećim zapadnim modelima kao što su Claude i ChatGPT. Model je premašio druge u zadacima kodiranja na prednjem kraju, prema platformi za procjenu umjetne inteligencije Arena. Vrijeme objavljivanja poklopilo se s govorom kineskog predsjednika Xi Jinpinga na Svjetskoj konferenciji o umjetnoj inteligenciji, naglašavajući rastuću konkurenciju između Kine i SAD-a u razvoju umjetne inteligencije. Unatoč američkim izvoznim ograničenjima, kineske tvrtke unapređuju svoje mogućnosti umjetne inteligencije kroz modele otvorenog koda i domaći hardver, kao što je Huawei Atlas 950 SuperPoD. Analitičari primjećuju sličnosti s prethodnim reakcijama na druga kineska izdanja umjetne inteligencije, što sugerira potencijalne koristi i izazove za zapadne tvrtke.
Procjena pristranosti (Sredina): Iako članak raspravlja o konkurentnoj dinamici između američke i kineske industrije umjetne inteligencije, predstavlja informacije iz više perspektiva, uključujući citate američkih analitičara i pominjanje američkih politika.
Zašto činjenice (85): This article reports on the release of Kimi K3 and its performance relative to U.S. models like Claude and ChatGPT, citing evaluations from Arena. It references the timing of the release in relation to President Xi's speech and mentions prior releases like Zhipu's GLM-5.2. The information is consist
Zašto objektivnost (70): The article presents the release of Kimi K3 as a significant development that challenges U.S. dominance, using emotionally charged language such as 'took the U.S. tech industry by surprise.' It emphasizes the impact on American companies and suggests a growing imbalance in the AI race, which reflect
Breitbart NewsNeovisanProgresivnoČinjenice 80Objektivnost 65prije 15 dana Izvještaj: Kineski 'Kimi K3' AI model odgovara performansama vrhunskih američkih sustavaNovi model umjetne inteligencije pod nazivom "Kimi K3" razvijen od strane kineske tvrtke Moonshot AI pokazao je performanse usporedive s vrhunskim američkim sustavima umjetne inteligencije kao što su Anthropic Fable 5 i OpenAI GPT-5.6 Sol, dok je znatno jeftiniji. Performanse modela iznenadile su promatrače industrije i izazvale zabrinutost u Silicijumskoj dolini zbog smanjenja tehnološkog jaza između Kine i Sjedinjenih Država. Testiranje od strane AI ocjenjivača Arena pokazalo je da je Kimi nadmašio nekoliko američkih modela u zadacima kodiranja i rangiranja teksta.
Procjena pristranosti (Progresivno): U članku se napredak kineske umjetne inteligencije opisuje kao značajan izazov za tehnološku dominaciju SAD-a, što sugerira potencijalne ekonomske i strateške implikacije.
Zašto činjenice (80): The article states that Kimi K3 matches U.S. models in performance while being more cost-effective, citing testing by Arena. It references earlier assessments by the U.S. government that suggested Chinese models were behind, but this is presented as outdated. The article also mentions the open-weigh
Zašto objektivnost (65): The tone is critical of U.S. technological advantage and highlights the implications of China's progress, suggesting a loss of U.S. leadership. The article uses terms like 'raising questions about the United States’ technological advantage' which implies a negative assessment. There is a clear focus
QuartzNeovisanSredinaČinjenice 80Objektivnost 60prije 15 danaAnthropic, startup za umjetnu inteligenciju, predložio je potencijalni ugovor od 10 milijardi dolara za zakup računalne snage umjetne inteligencije od Meta.
Procjena pristranosti (Sredina): Članak predstavlja sporazum kao poslovni prijedlog, bez da otvoreno favorizira bilo koju tvrtku ili izražava snažne ideološke sklonosti.
Zašto činjenice (80): The article accurately describes the $1.5 billion settlement and its significance, citing the legal process and the involvement of specific authors and attorneys. It references past rulings and the ongoing legal implications, which are consistent with other reports.
Zašto objektivnost (60): The article takes a critical stance, suggesting the settlement sets a bad precedent and could lead to more lawsuits. This introduces a degree of editorializing, implying negative consequences without providing counterarguments or balancing perspectives.
TechCrunchNeovisanSredinaČinjenice 30Objektivnost 65prije 11 dana SAD prijeti sankcijama protiv kineskih modela umjetne inteligencije zbog krađe intelektualne svojineAmerički ministar financija Scott Bessent najavio je da će administracija temeljito ispitati kineske modele AI otvorenog koda za potencijalnu krađu intelektualne svojine (IP), uz moguće sankcije protiv kineskih tvrtki ako se takva krađa potvrdi.
Procjena pristranosti (Sredina): U članku su predstavljene i akcije američke vlade i kontraargumenti vođa industrije i stručnjaka, nudeći više perspektiva o problemu krađe intelektualne svojine i destilacije modela.
Zašto činjenice (30): The article discusses Microsoft's partnership with Mistral, which is unrelated to the Microsoft earnings report. There is no connection to the primary source document, so factual alignment cannot be assessed.
Zašto objektivnost (65): The article presents the partnership in a neutral, factual tone.
★
Neka vijesti ostanu poštene.
ObjectiveNews financiraju čitatelji i bez oglasa je – pristranost vam pokazujemo, ne skrivamo. Podržite neovisno novinarstvo za 5 €/mjesec.
Postani podupiratelj
United States
