OpenAI’s safety lead defects to competitor Anthropic: Hashtag Trending for Friday, May 31st, 2024

Share post:

OpenAI’s Ex-Safety Lead Joins Rival Anthropic. Massive Ticketmaster Breach Fuels Speculation on Hacker Tactics. US Senators express “concern” over large Microsoft contract. An MIT professor throws cold water on economic growth predictions from AI advances.

These stories and more on this “here comes reality” edition of Hashtag Trending. I’m your host Jim Love, let’s get into it.

Jan Leike, the former head of OpenAI’s critical “superalignment” efforts to keep advanced AI systems aligned with human values, has left to join rival firm Anthropic.

Leike’s move comes after he leveled criticism at OpenAI following the company’s boardroom coup attempt to oust CEO Sam Altman last November over AI safety concerns.

At Anthropic, Leike will now lead a new oversight team working on scalable oversight, generalization of AI capabilities, and automated research into aligning AI with human interests.

Leike says, “My new team will work on scalable oversight, weak-to-strong generalization, and automated alignment research,”

The defection deals a blow to OpenAI’s reputation and feeds those who claim it is focusing more on commercialization through its Microsoft partnership and viral hit ChatGPT than responsible and secure AI development.

In contrast, Anthropic was founded by former OpenAI researchers emphasizing more responsible AI development. Backed by Amazon and Google, its Claude chatbot focuses on calculated, restrained responses.

OpenAI has promised to address Leike’s concerns and announced a new board-level safety committee on Tuesday, highlighting the priority AI labs are placing on responsible development amid the meteoric progress.

Sources include: Axios

A new research paper from MIT economist Daron Acemoglu offers a sobering counterpoint to optimistic forecasts that artificial intelligence will dramatically boost productivity and economic growth.

While many analysts have had predictions of huge growth in global GDP by as much as 100%.  Perhaps more realistic numbers from Goldman Sachs are in the  7% range over the next decade, Acemoglu argues the technology’s impact will be far more muted – estimating just a 0.5-0.9% GDP growth over 10 years.

[Quote] “AI will have implications for productivity, wages and inequality, but all of them are very hard to predict. This has not stopped a series of forecasts, often centering on the productivity gains that AI will trigger.” – Daron Acemoglu, MIT

Acemoglu contends previous waves of automation largely benefited business owners over workers. Using research from last year that estimated 20 percent of US workers could have half of their jobs done by an LLM, Acemoglu estimates that AI can save 27 percent in labor costs, or 14.4 percent in overall costs.

Those cost savings will of course come from job losses, which Acemoglu maintains will be primarily for low-wage workers.

And he calculates only modest productivity growth of 0.06% annually.

While not dismissing AI’s economic impact, Acemoglu argues the effects are far less transformative than trumpeted, at least in the short-term horizon.

For business leaders, the study injects a note of caution amid AI’s hype cycle. As powerful as the technology promises, prudent executives must weigh realistic productivity projections against implementation costs and societal impacts.

Sources include: The Register

In a massive data breach affecting over 560 million customers, hackers claim to have stolen a 1.3 terabyte database from Ticketmaster containing sensitive personal and financial information.  Ticketmaster remains tight-lipped on breach details, leading to speculation over how the threat actors gained access.

The high-profile breach marks the brazen resurfacing of hacker group BreachForums just days after the FBI’s second takedown of the notorious cybercrime forum and its Telegram channels.  The group was apparently looking for big target to relaunch it’s forum. Clearly, they found this with Ticketmaster and they are selling the stolen database on the relaunched BreachForums for $500,000.

The data reportedly includes customer names, addresses, emails, phone numbers, ticket purchase histories and partial payment card data dating back over a decade – the partial data includes the last four digits of the card number and expiry dates.

[Quote] “The attackers were able gain access to Ticketmaster’s AWS instances by pivoting from a managed service provider,” according to cybercrime monitor vx-underground’s analysis.

While unconfirmed, a suspected avenue of attack is via a third-party service provider, some reports talk about the attackers gaining access via a managed service provider who had access to Ticketmaster’s AWS instance.

For businesses, the incident underscores the persistent threats from criminal hacking groups and the importance of securing data, limiting service provider access, and having an incident response plan for breaches.

As Shane Huntley, Director of Google’s Threat Analysis Group, warned: “The cybercrime industry’s resilience continues to outpace disruption efforts.”

As damaging cyber attacks rapidly evolve, IT leaders must stay vigilant against supply chain compromises and develop robust safeguards to protect sensitive data.

Sources include: ITPro

US Senators Eric Schmitt and Ron Wyden have sent a letter to the Pentagon expressing “serious concern” over reported plans to invest heavily in upgrading to Microsoft’s product suite.

The bipartisan letter questions the Defense Department’s strategy of deepening its reliance on Microsoft instead of pursuing a multi-vendor approach for better competition, lower costs and improved cybersecurity outcomes.

[Quote]

“Through its buying power, DoD’s strategies and standards have the power to shape corporate strategies that result in more resilient cybersecurity services,” the senators wrote to Pentagon CIO John Sherman.

The concerns stem from a purported Pentagon memo requiring all offices to upgrade to Microsoft’s E5 licenses as part of its zero-trust security overhaul by next summer.

However, the senators argue this “doubles down” on a “failed strategy” after nation-state hackers exploited Microsoft software vulnerabilities in major breaches over the past year.

They are pressing Sherman on whether alternatives to an all-Microsoft solution were adequately evaluated, a sign of Washington’s eroding confidence in the tech giant’s cybersecurity capabilities.

For IT and security leaders, the high-stakes bureaucratic battle highlights the challenges of securing complex digital infrastructures and a realization that even giants like Microsoft are vulnerable to attack and compromise – making policymakers much more wary of vendor lock-in and supply chain risks.

Sources include:

And that’s it for today’s show. Remember that you can get us on Apple, Spotify or wherever you get your podcasts. We’re available on YouTube in both audio and video format.

Show notes are on Tech Newsday dot com or dot ca. Take your pick.

I’m your host Jim Love, have a thrilling Thursday.

 

 

SUBSCRIBE NOW

Related articles

Social Media Fraud Focuses Attacks On Truth Social: Cyber Security Today Weekend for January 18, 2025

Unmasking Social Media Scams: An Interview with Netcraft's Robert Duncan In this weekend edition of 'Cybersecurity Today,' host Jim...

Can Canada Get It’s Mojo Back? An Exclusive Interview With Jim Balsillie for Hashtag Trending

In this episode of the series, 'Can Canada Get Its Mojo Back?', host Jim Love explores the economic...

Microsoft Ends Support for Office 365 Apps on Windows 10: Hashtag Trending for Friday, January 17, 2025

Microsoft announces they won’t support  Office 365 on Windows 10, D-Wave achieves a quantum computing milestone, TikTok prepares...

Hackers Mount High Speed Microsoft 365 Attack: Cyber Security Today – January 17, 2025

Hackers exploit a high-speed Go library to target Microsoft 365 accounts worldwide, North Korea’s Lazarus group lures developers...

Become a member

New, Relevant Tech Stories. Our article selection is done by industry professionals. Our writers summarize them to give you the key takeaways