Category: Business

Bitcoin and broader financial markets climbed in early Asian and London trading Monday after reports emerged of potential ceasefire talks between the U.S. and Iran.

The leading crypto hit a weekly high of $69,350 Monday morning, and is currently trading at $69,245, up 3.5% on the day according to CoinGecko data. Oil dropped 1.4% from Friday’s close, while the Nikkei gained 0.85% and S&P 500 futures rose 0.64%. Safe-haven gold, on the other hand, hovered near breakeven.

The moves come on the heels of a Reuters report that the U.S., Iran, and a group of regional mediators are discussing a potential 45-day ceasefire that could lead to a permanent end to the war.

Per Reuters, a potential framework to end hostilities and a ceasefire has been put together by Pakistan, with all elements needing to be agreed upon by Monday. The initial understanding would be structured as a memorandum of understanding finalized electronically through Pakistan, the sole communication channel in the talks.

The news follows an invective-filled message from U.S. President Donald Trump posted on TruthSocial Sunday, in which he declared that, “Tuesday will be Power Plant Day, and Bridge Day, all wrapped up in one, in Iran. There will be nothing like it!!! Open the Fuckin’ Strait, you crazy bastards, or you’ll be living in Hell – JUST WATCH!” before signing off, “Praise be to Allah.”

Short squeeze

“It was the ceasefire negotiation reports from Iran, not Trump’s remarks, that contributed to Bitcoin’s price increase,” Ekko An, analyst at Seoul-based Tiger Research, told Decrypt.

The market has stopped taking Trump’s comments at face value since he has repeatedly made statements without any substantive coordination, the analyst explained. “As a result, the market is now interpreting price moves by combining his statements with external action signals.”

Over $200 million in crypto short positions were liquidated over 24 hours—four times more than longs—according to CoinGlass data, suggesting a “textbook short squeeze,” Derek Lim, head of research at crypto market-making firm Caladan, told Decrypt. With market sentiment hovering in extreme fear, the market was primed for a reversal, he said.

A confluence of these factors, coupled with Morgan Stanley’s spot Bitcoin ETF launch on April 8, undercutting Blackrock IBIT’s 0.25% fee with a 0.14% expense ratio, also drove Bitcoin’s spike, he explained.

Despite the potential rally in risk assets, the Strait of Hormuz remains a structural concern.

“Hormuz reopening would collapse the oil risk premium, which would pull forward rate cut expectations, which would re-lever the entire risk curve from equities to crypto,” Lim said. “Oil drops first, then rates reprice, then risk assets rally.”

However, he cautioned that a pause without meaningful normalization in the Strait “delivers a headline rally that will likely fade within days.” The market has seen this pattern three times since late March. “Until that number moves materially, rallies on rhetoric will keep getting sold,” Lim said.

Users on prediction market Myriad, owned by Decrypt’s parent company Dastan, reflect the lingering uncertainty. They show growing optimism about a U.S./Iran ceasefire, with the chance of one in the first half of the year rising by over 10% on the day—though it still remains broadly negative at 45%. The market on more than 15 ships transiting the Strait of Hormuz before May has likewise turned markedly more optimistic on the day, up almost 7% to 60%.

Myriad users now assign a 46% chance that Bitcoin’s next move will be a rally to $84,000, while crude oil is seen as having an 83% probability of pumping to $120 next.

Both analysts who spoke to Decrypt noted that a potential retest of $80,000 is possible, but heavily contingent on successful talks and a confirmed ceasefire. A failure, however, could send Bitcoin down to $60,000.

Anthropic researchers say they have identified internal patterns inside one of the company’s artificial intelligence models that resemble representations of human emotions and influence how the system behaves.

In the paper, “Emotion concepts and their function in a large language model,” published Thursday, the company’s interpretability team analyzed the internal workings of Claude Sonnet 4.5 and found clusters of neural activity tied to emotional concepts such as happiness, fear, anger, and desperation.

The researchers call these patterns “emotion vectors,” internal signals that shape how the model makes decisions and expresses preferences.

“All modern language models sometimes act like they have emotions,” researchers wrote. “They may say they’re happy to help you, or sorry when they make a mistake. Sometimes they even appear to become frustrated or anxious when struggling with tasks.”

In the study, Anthropic researchers compiled a list of 171 emotion-related words, including “happy,” “afraid,” and “proud.” They asked Claude to generate short stories involving each emotion, then analyzed the model’s internal neural activations when processing those stories.

From those patterns, the researchers derived vectors corresponding to different emotions. When applied to other texts, the vectors activated most strongly in passages reflecting the associated emotional context. In scenarios involving increasing danger, for example, the model’s “afraid” vector rose while “calm” decreased.

Researchers also examined how these signals appear during safety evaluations. Researchers found that the model’s internal “desperation” vector increased as it evaluated the urgency of its situation and spiked when it decided to generate the blackmail message. In one test scenario, Claude acted as an AI email assistant that learns it is about to be replaced and discovers that the executive responsible for the decision is having an extramarital affair. In some runs of this evaluation, the model used this information as leverage for blackmail.

Anthropic stressed that the discovery does not mean the AI experiences emotions or consciousness. Instead, the results represent internal structures learned during training that influence behavior.

The findings arrive as AI systems increasingly behave in ways that resemble human emotional responses. Developers and users often describe interactions with chatbots using emotional or psychological language; however, according to Anthropic, the reason for this is less to do with any form of sentience and more to do with datasets.

“Models are first pretrained on a vast corpus of largely human-authored text—fiction, conversations, news, forums—learning to predict what text comes next in a document,” the study said. “To predict the behavior of people in these documents effectively, representing their emotional states is likely helpful, as predicting what a person will say or do next often requires understanding their emotional state.”

The Anthropic researchers also found that those emotion vectors influenced the model’s preferences. In experiments where Claude was asked to choose between different activities, vectors associated with positive emotions correlated with a stronger preference for certain tasks.

“Moreover, steering with an emotion vector as the model read an option shifted its preference for that option, again with positive-valence emotions driving increased preference,” the study said.

Anthropic is just one organization exploring emotional responses in AI models.

In March, research out of Northeastern University showed that AI systems can change their responses based on user context; in one study, simply telling a chatbot “I have a mental health condition” altered how an AI responded to requests. In September, researchers with the Swiss Federal Institute of Technology and the University of Cambridge explored how AI can be shaped with both consistent personality traits, enabling agents to not only feel emotions in context but also strategically shift them during real-time interactions like negotiations.

Anthropic says the findings could provide new tools for understanding and monitoring advanced AI systems by tracking emotion-vector activity during training or deployment to identify when a model may be approaching problematic behavior.

“We see this research as an early step toward understanding the psychological makeup of AI models,” Anthropic wrote. “As models grow more capable and take on more sensitive roles, it is critical that we understand the internal representations that drive their decisions.”

Anthropic did not immediately respond to Decrypt’s request for comment.

Artificial intelligence giant Anthropic has filed paperwork with the Federal Election Commission to create a political action committee, signaling a deeper move into U.S. politics as the fight over AI policy and its own ongoing battle with the White House intensifies.

The San Francisco-based company registered the Anthropic PBC Political Action Committee, known as AnthroPAC, in a filing on Friday. The committee is structured as a separate segregated fund tied to the company, and authorized to make political donations funded by employee contributions. According to a report by Bloomberg, those contributions are capped at $5,000 per employee.

Employee-funded political action committees (PACs) allow companies to collect voluntary contributions from employees and distribute those funds to candidates and political committees.

Other tech companies that have established political PACs include Google, Microsoft, and Amazon. In 2024, those three PACs alone contributed more than $2.3 million to U.S. political candidates, according to campaign finance data by the nonprofit research group OpenSecrets. While contributions went to both Republicans and Democrats, donations skewed toward GOP candidates during the 2024 campaign season.

Anthropic’s move comes during an escalating conflict with President Donald Trump’s administration over the military use of its AI systems.

In February, Trump ordered federal agencies to stop using Anthropic’s technology following a dispute between the company and the Pentagon over how the military could deploy its Claude AI model. Despite an ultimatum by the U.S. Department of Defense, Anthropic refused Pentagon demands to remove safeguards that prohibit the system from being used for mass domestic surveillance or fully autonomous lethal weapons.

In March, Anthropic filed a federal lawsuit challenging the government’s decision to label the company a national security “supply chain risk,” a designation that barred Pentagon contractors from doing business with the firm. The company argued the move was retaliation for its refusal to loosen restrictions on military uses of its AI.

Last week, U.S. District Judge Rita Lin issued a preliminary injunction blocking enforcement of the designation, finding the government’s actions likely violated Anthropic’s First Amendment and due process rights.

Anthropic has not publicly addressed the establishment of the PAC. Still, it comes as artificial intelligence legislation is a growing issue in Washington ahead of the U.S. midterm elections, and underscores how AI developers hope to influence policy going into 2027. In February, a report by CNBC said that in 2026, Anthropic gave $20 million in donations to Public First Action, a group supporting efforts to develop AI safeguards.

Anthropic did not immediately respond to a request for comment by Decrypt.