Connect with us
DAPA Banner

Crypto World

OpenAI launches paid Safety Fellowship

Published

on

OpenAI launches paid Safety Fellowship

The AI news out of OpenAI this week has a sharp edge: the company launched a paid Safety Fellowship offering $3,850 weekly stipends to external researchers studying what could go wrong with advanced AI — announced within hours of a New Yorker investigation reporting that OpenAI had dissolved its internal safety teams and quietly removed the word “safely” from its IRS mission statement.

Summary

  • The OpenAI Safety Fellowship, announced April 6, runs from September 14, 2026 through February 5, 2027; fellows receive a $3,850 weekly stipend, approximately $15,000 in monthly compute resources, and mentorship from OpenAI researchers, but will not have access to the company’s internal systems
  • Priority research areas include safety evaluation, ethics, robustness, scalable mitigations, privacy-preserving methods, agentic oversight, and high-severity misuse — applications close May 3, with fellows notified by July 25
  • The New Yorker’s Ronan Farrow reported the same week that OpenAI had dissolved its superalignment team, its AGI Readiness team, and its Mission Alignment team since 2024, and that an OpenAI representative responded to a journalist asking about existential safety researchers with: “What do you mean by existential safety? That’s not, like, a thing.”

OpenAI announced the fellowship on April 6 as “a pilot program to support independent safety and alignment research and develop the next generation of talent.” The program pays $3,850 per week, over $200,000 annualized, plus roughly $15,000 in monthly compute and mentorship from OpenAI researchers. Fellows work from Constellation’s Berkeley workspace or remotely, and applications close May 3. The fellowship is not limited to AI specialists — OpenAI is recruiting from cybersecurity, social science, and human-computer interaction alongside computer science.

The timing is the story. Ronan Farrow’s investigation in The New Yorker, published the same day, documented that OpenAI had dissolved three consecutive internal safety organizations over 22 months. The superalignment team was shut down in May 2024 after co-leads Ilya Sutskever and Jan Leike departed. Leike wrote on his way out that “safety culture and processes have taken a backseat to shiny products.” The AGI Readiness team followed in October 2024. The Mission Alignment team was disbanded in February 2026 after just 16 months. The New Yorker also reported that when a journalist asked to speak with OpenAI’s existential safety researchers, a company representative replied: “What do you mean by existential safety? That’s not, like, a thing.”

Advertisement

The fellowship explicitly does not replace internal infrastructure. Fellows receive API credits and compute resources but no system access, positioning the program as arm’s-length research funding rather than a rebuild of the dissolved teams.

What the Fellowship Requires Fellows to Produce

The research agenda spans seven priority areas: safety evaluation, ethics, robustness, scalable mitigations, privacy-preserving safety methods, agentic oversight, and high-severity misuse domains. By the program’s end in February 2027, each fellow must produce a substantive output — a paper, benchmark, or dataset. Specific academic credentials are not required; OpenAI stated it prioritizes research ability, technical judgment, and execution capacity.

Why This Matters Beyond the AI Industry

As crypto.news has reported, confidence in frontier AI companies’ stated safety commitments is a market signal that affects capital allocation across AI infrastructure, AI tokens, and the DePIN and AI agent protocols sitting at the intersection of crypto and artificial intelligence. As crypto.news has noted, OpenAI’s spending trajectory and the credibility of its operational priorities are tracked closely by investors evaluating the AI infrastructure sector — a sector with growing overlap with blockchain-based systems. Whether external fellows working without internal access can meaningfully influence model development is a question the first cohort’s research will begin to answer in early 2027.

Advertisement

Source link

Continue Reading
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Crypto World

Epic Market Flash Crash Killed Bull Market: Is Crypto Healthier Now?

Published

on

Epic Market Flash Crash Killed Bull Market: Is Crypto Healthier Now?

Key takeaways:

  • Bitcoin orderbook depth has plummeted by 50% since September 2025, signaling a substantial decline in overall market liquidity.

  • Indicators suggest that the current market fragility stems more from recent 2026 trends than from the 2025 flash crash itself.

Bitcoin (BTC) and crypto markets took a massive hit on Oct. 10, 2025, precisely 6 months ago. That devastating flash crash wiped out a record-breaking $19 billion in leveraged positions while some altcoins collapsed 40% to 80%. Many traders speculated that multiple market makers had been wiped out, while others accused the Binance exchange of blatant manipulation.

Was the crypto market structure actually altered after the October 2025 crash, and what has changed in liquidity, derivatives markets, and institutional metrics?

Aggregate Bitcoin spot +1% to -1% orderbook depth, USD. Source: CoinAnk

Bitcoin’s aggregate orderbook depth, ranging from +1% to -1%, typically oscillated between $180 million and $260 million in September 2025. On most days, there would be a healthy $90 million in bids, but that was not the case on Oct. 10, 2025. A mix of technical issues at Binance and auto-deleveraging on decentralized exchanges caused a temporary liquidity lapse.

During the flash crash, Bitcoin’s orderbook depth entered a downward spiral, stabilizing near $150 million by mid-November 2025. Currently, Bitcoin’s order book depth seldom exceeds $130 million, down 50% from levels seen in September 2025.

Advertisement

The already fragile market conditions deteriorated further in February 2026. Bitcoin’s orderbook depth plunged below $60 million for nearly 10 days as the price struggled to hold the $65,000 level. Cryptocurrency market volumes declined considerably, especially in the derivatives markets.

Total crypto trading volume, USD. Source: TokenInsight

Cryptocurrency derivatives volumes oscillated between $40 billion and $130 billion over the past 30 days, falling short of the $200 billion mark commonly seen in September 2025. Still, the reduced appetite for futures contracts is not necessarily a bearish indicator as longs (buyers) and shorts (sellers) are evenly matched at all times.

Demand for bullish leverage remains weak, ETF volumes lag

The Bitcoin perpetual futures funding rate can be used to assess traders’ risk appetite.

Bitcoin perpetual futures annualized funding rate. Source: Laevitas

Under normal conditions, the indicator should range between 6% to 12% to compensate for the cost of capital. Excessive demand for bearish leverage can push the indicator below 0%, meaning shorts are the ones paying to keep their positions open. Data indicate stable conditions throughout November 2025, followed by a sharp decline in February 2026.

Curiously, volumes of US-listed spot Bitcoin exchange-traded funds (ETFs) were not impacted by the Oct. 10, 2025 flash crash. In fact, by late November, activity in those instruments jumped to their highest levels in 20 months at $11.5 billion per day. 

Related: Binance adds spot trading guardrails to limit abnormal executions

Advertisement
US-listed spot Bitcoin ETFs daily trading volume, USD. Source: Coinglass

Bitcoin ETFs regularly traded at volumes above $4 billion per day between January and March 2026, but eventually fell below $3.3 billion by the first week of April. Similarly, US-listed Ether (ETH) ETFs average daily volume dropped to $1 billion, down from $2 billion in September 2025. 

Orderbook depth, funding rate, derivatives and ETF volumes all point to a much less healthy cryptocurrency market in April 2026 relative to 6 months prior. However, given that the market structure held relatively firm through February 2026, the relevance of the Oct. 10, 2025 flash crash seems much less than previously imagined.