Month: June 2024

Enlarge / Stephen Ubl, president and chief executive officer of Pharmaceutical Research and Manufacturers of America (PhRMA), speaks during a Bloomberg Live discussion in Washington, DC, in 2017. (credit: Getty | Andrew Harrer)

After a series of decisive court losses, the pharmaceutical industry appears to be taking its fight against Medicare drug price negotiations directly to the people—and the White House is not impressed.

This week, the high-powered industry group PhRMA (the Pharmaceutical Research and Manufacturers of America) released two eye-catching attacks on federal efforts to lower America's singularly astronomical drug prices. In a press release Tuesday, PhRMA announced an analysis suggesting that the Medicare drug price negotiations—part of the Biden administration's 2022 Inflation Reduction Act—could actually cost some seniors and people with disabilities slightly more in out-of-pocket costs. The analysis, however, relies on a key—and questionable—assumption that the federal government will set price limits using the highest possible estimate for maximum fair prices in 2026.

Milliman, the consulting firm PhRMA commissioned to do the study, cautioned that the actual prices "will certainly vary due to differences in unit cost and utilization trend, 2026 benefit designs, and actual 2026 maximum fair prices."

Enlarge / Ozone-producing chemicals come from a variety of sources and don't respect state borders. (credit: John Edward Linden)

On Tuesday, a slim majority of the US Supreme Court issued an emergency ruling that places a stay on rules developed by the Environmental Protection Agency, meant to limit the spread of ozone-generating pollutants across state lines. Because it was handled on an emergency basis, the decision was made without any evidence gathered during lower court proceedings. As a result, the justices don't even agree on the nature of the regulations the EPA has proposed, leading to a blistering dissent from Justice Amy Coney Barrett, who was joined by the court's three liberal justices.

Bad neighbors

The rule at issue arose from the EPA's regular process of revisiting existing limits in light of changes in public health information and pollution-control technology. In this case, the focus was on ozone-producing chemicals; in 2015, the EPA chose to lower the limit on ozone from 75 to 70 parts per billion.

Once these standards are set, states are required to submit plans that fulfill two purposes. One is to limit pollution within the state itself; the second involves pollution controls that will limit the exposure in states that are downwind of the pollution sources. The EPA is required to evaluate these plans; if they are deemed insufficient, the EPA can require the states to follow a federal plan devised by the EPA.

Enlarge / The logo for PaLM 2, a Google large language model. (credit: Google)

Google announced today that it has added support for 110 new languages to Google Translate, nearly doubling the number of languages that can be translated.

The company used the PaLM 2 large language model to facilitate these additions.

In a blog post, Google Senior Software Engineer Isaac Caswell claimed that the newly added languages are spoken by more than 614 million people, or about 8 percent of the global population.

Enlarge / An illustration created by OpenAI. (credit: OpenAI)

On Thursday, OpenAI researchers unveiled CriticGPT, a new AI model designed to identify mistakes in code generated by ChatGPT. It aims to enhance the process of making AI systems behave in ways humans want (called "alignment") through Reinforcement Learning from Human Feedback (RLHF), which helps human reviewers make large language model (LLM) outputs more accurate.

As outlined in a new research paper called "LLM Critics Help Catch LLM Bugs," OpenAI created CriticGPT to act as an AI assistant to human trainers who review programming code generated by the ChatGPT AI assistant. CriticGPT—based on the GPT-4 family of LLMS—analyzes the code and points out potential errors, making it easier for humans to spot mistakes that might otherwise go unnoticed. The researchers trained CriticGPT on a dataset of code samples with intentionally inserted bugs, teaching it to recognize and flag various coding errors.

The development of CriticGPT involved training the model on a large number of inputs containing deliberately inserted mistakes. Human trainers were asked to modify code written by ChatGPT, introducing errors and then providing example feedback as if they had discovered these bugs. This process allowed the model to learn how to identify and critique various types of coding errors.

Enlarge (credit: Getty Images)

Mac malware that steals passwords, cryptocurrency wallets, and other sensitive data has been spotted circulating through Google ads, making it at least the second time in as many months the widely used ad platform has been abused to infect web surfers.

The latest ads, found by security firm Malwarebytes on Monday, promote Mac versions of Arc, an unconventional browser that became generally available for the macOS platform last July. The listing promises users a “calmer, more personal” experience that includes less clutter and distractions, a marketing message that mimics the one communicated by The Browser Company, the startup maker of Arc.

When verified isn’t verified

According to Malwarebytes, clicking on the ads redirected web surfers to arc-download[.]com, a completely fake Arc browser page that looks nearly identical to the real one.