Month: August 2024

In a letter to Delta, the software developer CrowdStrike accused the airline of causing the extensive impact of the recent incident, with Microsoft sending a similar message, blaming Delta for using outdated IT solutions. Delta, however, maintains that it bears no responsibility for the malfunctioning software, let alone for the failures of other systems worldwide. CrowdStrike now also faces the threat of a class-action lawsuit from investors.

The disputes have revealed some interesting details. By the time CrowdStrike’s CEO reached out to Delta’s CEO (once, late on the night of July 22), nearly four days had passed since the incident, and Delta had already restored its critical systems and the operations of most of its other computers. According to Delta, the offer of assistance came “too late,” and the phone call was “useless and untimely.” Delta denies that the software developer was “tirelessly” working to restore Delta’s systems, as claimed.

Furthermore, it is alleged that after Windows computers started shutting down globally, CrowdStrike did not take urgent measures or properly assess the scale and consequences of the failure. The only help provided in the first 65 hours after the incident was a link to a public website advising a manual reboot of affected PCs and the removal of faulty files. Worse, the update for automated problem resolution released on July 21 contained an additional bug, causing many machines to require further intervention before they could function again.

Delta claims to have invested billions of dollars in IT solutions, and the prolonged restoration of its IT systems was due to the company’s heavy reliance on CrowdStrike and Microsoft. Specifically, it was reported that about 60% of Delta’s critical applications and related data, including reservation systems, depend on Windows and Falcon. This trust in the software manufacturers backfired.

In response, CrowdStrike stated that Delta is promoting a misleading narrative. The company claims that CEO George Kurtz called a Delta representative within four hours of the incident on July 19, and CrowdStrike’s chief security officer established direct contact with Delta colleagues within hours of the incident. Both companies’ teams were closely collaborating just hours after the incident, with CrowdStrike providing far more support than just a link to a site with basic information.

As indirect evidence, one of Delta’s board members posted on LinkedIn, praising CrowdStrike’s CEO and his team for doing incredible work, tirelessly working under difficult conditions to resolve the issues. Delta declined further comment, but after expressing gratitude, it will undoubtedly be more challenging to take legal action against those responsible for the incident. In its letter, CrowdStrike once again urged Delta to stop trying to evade responsibility and to inform customers and shareholders about everything known regarding the incident, the decisions made, and the actions taken.

Anthropic has announced the launch of an expanded bug bounty program, offering external cybersecurity specialists rewards of up to $15,000 for identifying critical issues in its artificial intelligence systems.

This initiative is aimed at uncovering “universal bypass methods,” which are potential exploits that could consistently circumvent AI security measures in high-risk areas such as chemical, biological, radiological, and nuclear threats, as well as in cyberspace. According to VentureBeat, Anthropic will invite ethical hackers to test its system even before its public release, with the goal of preventing potential exploits that could lead to the misuse of its AI systems.

Interestingly, this approach differs from the strategies of other major AI players. For example, while OpenAI and Google also have bug bounty programs, they are more focused on traditional software vulnerabilities rather than AI-specific exploits. Additionally, Meta✴ recently faced criticism for its relatively opaque stance on AI security research. In contrast, Anthropic’s clear emphasis on transparency sets a new standard in this area.

However, the effectiveness of bug bounty programs in addressing the full spectrum of AI security issues remains debated. Experts suggest that a more comprehensive approach may be needed, involving extensive testing, improved interpretability, and possibly new governance frameworks to ensure that AI systems globally align with human values.

The program will initially launch as an invitation-only initiative (closed testing) in partnership with the well-known platform HackerOne. In the future, Anthropic plans to expand the program, making it open and creating a separate independent model for industry collaboration on AI security.

The surge in the electric vehicle (EV) market in China is widely regarded as the result of a deliberate government policy that subsidized not only manufacturers of such vehicles but also consumers. Even in uncertain economic conditions, New Energy Vehicles (NEVs) continue to expand in China, with their market share surpassing 50% for the first time in July.

This information comes from CnEVPost, citing the China Passenger Car Association (CPCA). In July, hybrids and electric vehicles accounted for 51.1% of the primary car market in China. Essentially, vehicles powered solely by internal combustion engines (ICE) were in the minority among those sold on the Chinese market last month. In numerical terms, 878,000 hybrids and battery electric vehicles were sold in China in July, second only to the record set in December of last year, when sales reached 947,347 units. The year-over-year sales growth of hybrids and electric vehicles reached 36.9%, with a sequential increase of 2.8%.

The term NEV, as used by Chinese statistics authorities, includes not only hybrids with internal combustion engines and battery electric vehicles but also hydrogen fuel cell vehicles, though their sales volumes are negligible. In July, 54.8% of NEV sales in China were battery electric vehicles, amounting to 482,000 units. This represents a 14.3% increase from the previous year, although it is a 1.6% decrease from June.

The robust growth in NEV sales can largely be attributed to plug-in hybrids, with 396,000 units sold in July. This marks an 80.4% increase from July of the previous year, with a sequential growth of 8.8%. Plug-in hybrids accounted for 45.1% of NEV sales in China in July. The statistics do not break down hybrids by subtype, though they also vary by powertrain configuration.

In terms of the overall dynamics of passenger vehicle sales in China in July, a total of 1.72 million crossovers, sedans, and minivans were sold, representing a 2.8% decrease compared to the same period last year, and a sequential decline of 2.6%. Returning to the NEV category, their market share in China was only 36.1% in July of last year, and 48.4% in June of this year.

Among Chinese manufacturers, NEVs accounted for an impressive 73.9% of sales in July, while in the luxury segment overall, the figure did not exceed 27%. Foreign automakers with joint ventures in China offered hybrids and electric vehicles in only 8.3% of cases in July. NEV exports from China in July increased by 1.3% year-over-year to 92,000 units, with a sequential growth of 18.2%. Overall, nearly one in four new vehicles exported from China in July was either a hybrid or an electric vehicle, with the latter accounting for 72.7% of exports in July.

Today, Intel confirmed that its plans to launch Arrow Lake processors remain “unchanged,” with deliveries set to begin “later this year.” The Intel Innovation event, originally scheduled for September 24, 2024, will be postponed to 2025 to save costs due to the company’s losses in the second quarter of 2024. After news of the Intel Innovation 2024 event postponement surfaced, many enthusiasts and analysts speculated about a possible delay in the release of the Arrow Lake processor family. This speculation was based on Intel’s plans to unveil the new chip at the Innovation 2024 event. However, today, an Intel representative stated that these rumors are not true.

“Intel’s launch plans, timelines, or product readiness have not changed. We are excited about the new product announcements and the momentum we have, including important announcements this fall. We will share more details about our next-generation desktop processors codenamed Arrow Lake later this year,” Intel stated. The company promised to make “important announcements” this fall. The Lunar Lake processors are expected to be showcased at the IFA 2024 trade show in Germany early next month. There are also rumors about a possible announcement of the Battlemage generation desktop graphics processors.

The Arrow Lake family is particularly intriguing for the desktop segment as it will bring an entirely new architecture, be more energy-efficient than current chips, and promise a significant boost in performance.

During Asian trading, Bitcoin (BTC) briefly surpassed the $62,000 mark amidst a broader market rally, recovering losses from a sharp drop earlier in the week. The liquidation of short futures positions (bearish bets) worth nearly $100 million, positive stock market trends, and expectations of Bitcoin repeating its historical cycles were key factors driving this surge.

Over the past 24 hours, Bitcoin rose by 7.2%, marking one of its largest single-day gains in recent months. Some analysts attribute this rise to the positive sentiment in the stock market and expectations that Bitcoin will continue to follow its historical market cycles. “Now that the Bank of Japan has stated that it will not raise interest rates, and Jump Trading is about to exhaust its coin reserves for sale, I don’t see prices below $50,000 (except for brief fluctuations), and this may never happen again,” Michael Terpin, founder of Transform Ventures, said in his letter to CoinDesk on Friday.

He added that “regardless of the next 60 days, the bull market will continue to follow traditional four-year cycles with solid gains in October and November.” There was also speculation that if Donald Trump were to win, a new influx of buyers could drive Bitcoin’s price above $100,000. Moreover, within six months after a halving, pullbacks are typical, and the current fifth Bitcoin cycle is no exception. “October and November are historically strong months for Bitcoin, especially in the year of the halving and the year following it,” Terpin added.

Bitcoin’s rise also contributed to the price increase of other major tokens. Ethereum (ETH) and Toncoin (TON) rose by 10%, while SOL and ADA increased by 5%. The CoinDesk 20 (CD20) index, which tracks the largest tokens by market capitalization, increased by 5.35%.

American markets also saw gains the day before: the S&P 500 index had its best day since November 2022, and the Nasdaq 100 index rose by 3.1%, helping to offset losses from Monday’s drop, which had caused significant losses in both stock indices and cryptocurrencies.

Perplexity AI, a company based in San Francisco and founded by former Google intern Aravind Srinivas just three months before the launch of ChatGPT, has developed a search engine powered by artificial intelligence that generates real-time answers to user queries based on information from the internet. Instead of developing its own model, Perplexity opted to license a combination of AI systems from various providers, such as OpenAI and Google. Initially, the Perplexity search engine used a licensed version of Microsoft’s Bing index but later abandoned it in favor of its own index and ranking system. “We use signals from all kinds of search engines, but we have our own crawler and ranking system,” claims Perplexity’s business director, Dmitry Shevelenko.

This year, the company successfully raised $250 million in investments, tripling its estimated value from $1 billion in April to $3 billion today. Perplexity’s investors include Nvidia, Amazon founder Jeff Bezos, OpenAI co-founder Andrej Karpathy, and Meta’s chief AI scientist Yann LeCun.

Perplexity’s growth comes amid the integration of AI features into Google’s search engine and the launch of OpenAI’s SearchGPT prototype. “In the end, a smaller player in this field has two advantages: speed and focus,” says Shevelenko. “Our users and team think of only one thing when it comes to Perplexity: the place where you get answers to your questions. Competition sharpens our focus even further.”

At the beginning of the year, based on extrapolated monthly sales, Perplexity’s annual revenue was estimated at $5 million, but this projection has now increased sevenfold to $35 million. The company is now transitioning from a subscription-based business model to monetizing through paid advertising. This shift brings Perplexity closer to direct competition with Google, which dominates the $300 billion advertising industry.

The platform targets journalism and academic circles as sources of reliable information and data. Perplexity views this focus as an advantage over traditional search engines that index a much broader range of websites. However, some analysts believe that the introduction of advertising could negatively impact user growth due to concerns about a “less trustworthy environment” and skepticism about the impartiality of search results.

“For Perplexity to become a useful product on the open internet, there need to be good business models for publishing new and updated facts about the world,” says Shevelenko. “If you want to align incentives [with journalism] in the long term, revenue sharing is a more effective way to do that than one-time lump-sum payments, which OpenAI has adopted.”

According to Shevelenko, the company will share a “double-digit” percentage of its revenue with each partner news publisher. He mentioned that agreements have already been made with Time, Der Spiegel, and Fortune, and more than 50 publishers expressed interest in joining Perplexity’s revenue-sharing program within two weeks of its launch. It’s worth noting that Forbes and Wired previously accused Perplexity of intellectual property violations, “reproducing stories without clear attribution and copying websites that explicitly blocked its crawlers.” The company took the criticism seriously, making changes to the user interface to make citations more visible and altering the format of aggregated answers.

Many experts believe that the AI-based search market is “heating up.” They argue that general web search is becoming redundant due to new ways of matching users with information, products, and services. The risk, however, lies in whether these new methods will be reliable enough for widespread use. “AI stubbornly tends toward confabulation,” says analyst Joseph Teasdale. “At the scale of billions of queries per day, serious errors are inevitable.”