Having a health insurance plan with a high deductible could not only cost you—it could also kill you.

A new study in JAMA Network Open found that people who faced those high out-of-pocket costs as well as a cancer diagnosis had worse overall survival and cancer-specific survival than those with more standard health plans.

The findings, while perhaps not surprising, are a stark reminder of the fraught decisions Americans face as the price of health care only continues to rise and more people try to offset costs by accepting insurance plans with higher deductibles—that is, higher out-of-pocket costs they have to pay before their health insurance provider starts paying its share.

The issue is particularly critical right now for people who have insurance plans through the Affordable Care Act marketplace. Prices for those plans have skyrocketed this year after Congress failed to extend critical tax credits. Without those credits, monthly premiums for ACA plans have, on average, more than doubled. Early data on ACA enrollments for 2026 not only suggests that fewer people are signing up for the plans, but also that those who are enrolling are often choosing bronze plans, which are high-deductible plans.

In the study, researchers considered plans to be "high-deductible health plans" (HDHPs) if their deductibles were at least $1,200 to $1,350 for individuals or $2,400 to $2,700 for families between 2011 and 2018 (with the cutoffs increasing within the ranges during that time). For context, the average individual deductible for an ACA bronze plan in 2026 is about $7,500, according to KFF.

Risky plans

Based on previous data, such high out-of-pocket costs are known to lead people to delay or decrease health care—they may skip doctor visits, put off diagnostics, and avoid treatments. But for the new study, researchers led by Justin Barnes at Mayo Clinic in Rochester, Minnesota, wanted to know, more directly, if the plans were linked to lower survival—specifically for cancer patients, who obviously need more care than others.

The study used health survey data from a nationally representative group of over 147,000 people. In all, nearly 9,800 had been diagnosed with cancer, and of those, around 2,300 had HDHPs. Another group of over 37,000 who had not been diagnosed with cancer also had HDHPs. Adjusting for demographic factors and health status, cancer patients on an HDHP had a lower overall survival rate (46 percent higher risk of death) and lower cancer-specific survival rate (34 percent higher risk of death) than people on standard-deductible plans. In further analyses, the researchers found that, for cancer patients, whether their HDHP had a health savings account (HSA) did not make a difference. In survey responses, cancer patients on HDHPs reported more financial worry than people on standard plans.

By contrast, people on an HDHP who did not have a cancer diagnosis did not have lower survival rates than people on other plans, suggesting that the plans might not affect your survival as long as you don't have any serious health conditions, such as cancer.

Barnes and his colleagues conclude, "In the current political environment, in which there may be proliferation of HDHPs, these results suggest that HDHP proliferation could exacerbate adverse cancer outcomes. These data underscore the importance of educating patients with cancer about potential (increased) need for health care utilization, educating the public about potential risks of HDHPs, identifying policy solutions to decrease costs of care, and optimizing health policies to avoid disincentives for seeking what may turn out to be necessary and lifesaving or life-prolonging care."

Read full article

Comments

LG has launched a subscription program in the UK that allows people to make monthly payments in order to rent LG TVs, soundbars, monitors, and speakers.

LG Flex customers can sign up for one-, two-, or three-year subscriptions to get lower monthly payments.

“At the end of your subscription, you can apply for a free upgrade, keep paying monthly, or return your device,” the LG Flex website says. Subscribers will have to pay a £50 (about $69) fee for a “full removal service,” including dismounting and packaging, of rental TVs.

LG also claims on its website that it won’t penalize customers for “obvious signs of use, such as some scratching, small dents, or changes in the paintwork.” However, if you damage the rental device, LG “may charge you for the cost of repair as outlined by the Repair Charges set out in your agreement.” LG’s subscription partner, Raylo, also sells insurance for coverage against “accidental damage, loss, and theft" of rented devices.

As of this writing, you can buy LG’s 83-inch OLED B5 2025 TV on LG’s UK website for £2,550 (about $3,515). Monthly rental prices range from £93 ($128), if you commit to a three-year-long rental period, to £277 ($382), if you only commit to a one-month rental period. Under the three-year plan, you can rent the TV for 27 months before you end up paying more to rent the TV than you would have to own it. At the highest rate, your rental payments will surpass MSRP after nine months.

The cheapest item available through LG Flex as of this writing is LG’s 3.1.1 channel soundbar with Dolby Atmos from 2024. You can buy it from LG’s UK website for £600 ($827) or rent it for £22 ($30) to £76 ($105) per month.

LG Flex can grant people instant access to electronics they may not be able to pay for currently. There are other options, like interest-free payment programs (which LG also offers), but those programs can require larger monthly payments than rentals do.

Still, renting gadgets from LG won’t make financial sense for many. The program can be handy for people, businesses, and other entities that require high-end gadgets for a limited amount of time or that want frequent upgrades. But renting electronics quickly becomes more costly than buying something. It’s often more prudent to save money that would have been used for monthly payments and instead invest it in something that you can own and maintain for years.

LG's rental gadgets are only available in the UK. Ars Technica reached out to LG about whether it has plans to rent electronics in the US and will update this article if we hear back. As it stands, the program adds to the growing number of tech companies encouraging people to rent, instead of buy, their products. HP, for instance, has a printer rental program, and NZXT launched a gaming desktop rental program in 2024. Logitech CEO Hanneke Faber has also floated the idea of a subscription-based mouse. 

LG Flex is just one example of how the South Korea-headquartered company is finding creative ways to build recurring revenue instead of relying solely on sales of its expensive gadgets, as such purchases occur sparingly and less predictably. Last year, LG launched a similar rental program for home appliances in Singapore and Malaysia.

Read full article

Comments

Federal immigration agents stationed in Minneapolis need not read polls or confront protesters to know how the city feels about their presence. Walking around just about any neighborhood in the area lately should provide a glimpse into the vast sprawl of graffiti and homemade yard signs expressing residents’ bone-deep aversion to ICE. One poster in particular, though, has been increasingly decorating the storefront windows of local restaurants, coffee shops, yarn stores, pubs, and bowling alleys, urging in no uncertain terms: “ICE out of Minneapolis.”

This sign seems to have struck a chord within the community, not just because of its blunt message but the form it’s riffing on: a familiar red municipal sign highlighting snow emergency routes, already strewn throughout streets in the Twin Cities. While the original evokes the grill of a snowplow truck clearing out roads in the wake of a blizzard, the anti-ICE version includes helmets, rifles and handcuffs in the slushy waste. The new sign’s growing popularity suggests it’s tapping into residents’ regional identity as much as it is their love of creative protest art.

View this post on Instagram

Reimagining a local icon

Burlesque of North America, a local design studio specializing in graphic arts and screenprinting, created the sign as a response to ICE’s incursion into Minneapolis. Owners Mike Davis and Wes Winship had previously created an anti-ICE enamel pin back in November when they first got the eerie sense something like Operation Metro Surge lay on the horizon.

After their friends who run the nearby restaurant Hola Arepa were targeted by ICE in early December, the Burlesque team began playing with ideas for a protest poster. It didn’t take long for them to arrive at a concept rooted in the transportation department’s snow emergency sign.

“We’ve been figuring out how to handle literal ice here for centuries,” Winship says. “And we’ve got this sign that’s pre-built, alerting people: There’s an emergency and we need to remove frozen precipitation from the streets.”

It was clean, crisp iconography, on which to project a message of resistance. On top of that, it was instantly recognizable.

“For the locals, everyone knows the sign. Everyone’s been living with it and responding to it,” says Davis. “But even people from out of town who don’t know the reference, they can still tell what it means and connect with it.”

After working on a mockup, the two paused on the project as the holidays kicked into high gear. It remained set aside until January 7, the day an ICE agent shot and killed Renee Nicole Good. That drastic escalation spurred the pair back into action on their poster. By the following night, Davis and Winship had completed all the design elements. The next day, they’d screenprinted the first of what would become over 5,000 copies to distribute around Minneapolis. They also offered free PDFs on their website, an open-source touch allowing out-of-towners to print signs and shirts of their own.

The power of posters

Although projects like concert posters are Burlesque’s bread and butter, the team has been rooted in socially conscious art for ages. Amid the Syrian refugee crisis, for instance, they created a Refugees Welcome image that went viral in 2015 when the Walker Art Center featured it and several news outlets ultimately wrote about it. After George Floyd was killed in 2020, Davis and Winship printed out several artists’ protest graphics for free and handed out copies from their delivery truck at the intersection quickly dubbed George Floyd Square. 

It’s in that very area where the city’s love of protest art is most evident—made manifest by the iconic raised fist of steel in Floyd’s memory, designed by artist Jerome Powell-Karis.

The Burlesque team have now seen, once again, how much that love still radiates throughout the city during moments of upheaval. As they handed out their anti-ICE posters, restaurant and shop owners lit up at the sight of it. Davis brought 600 copies to a protest in Powderhorn Park the weekend after Good was killed, and his supply was picked clean within 15 minutes. When the team later asked for donations to cover their ink and paper costs, they hit their goal in three days. Meanwhile, the response on Instagram was unlike anything the pair had ever experienced. 

“We’re still being contacted almost hourly by someone who either wants one or wants an entire stack to give away at their bike shop or trivia night,” Davis says.

The signs seem especially resonant at restaurants, where some owners are now opting to keep their doors locked during daytime hours to prevent ICE from entering. Local service workers seem haunted by recent reports of ICE agents eating at a restaurant in a town outside of Minneapolis, only to detain several employees afterward. The anti-ICE signs add some extra oomph to the locked doors, and let prospective diners know where the establishment stands. (Many restaurants without this particular sign have homemade anti-ICE signage of their own.)

Although the snow emergency poster has started gaining a lot of traction, it feels perfectly at home among all the other protest art on display throughout the city—whether it’s the signs featuring Minnesota state bird, the loon, melting ice with laser eyes, an image with roots in local lore; all the inventive DIY anti-ICE entries at last weekend’s Art Sled Rally; or the ubiquitous stickers depicting melted ice. The Twin Cities community has clearly come together around its resistance to the current siege and Burlesque is happy to have helped play some small part in it.

“I’m not going to be out there with a gas mask at the Federal Building, catching rubber bullets,” Davis says. “So, it feels good that I have something to contribute to the cause. Like, I’m a graphic designer. This is what I can do.”

On Friday, OpenAI engineer Michael Bolin published a detailed technical breakdown of how the company's Codex CLI coding agent works internally, offering developers insight into AI coding tools that can write code, run tests, and fix bugs with human supervision. It complements our article in December on how AI agents work by filling in technical details on how OpenAI implements its "agentic loop."

AI coding agents are having something of a "ChatGPT moment," where Claude Code with Opus 4.5 and Codex with GPT-5.2 have reached a new level of usefulness for rapidly coding up prototypes, interfaces, and churning out boilerplate code. The timing of OpenAI's post details the design philosophy behind Codex just as AI agents are becoming more practical tools for everyday work.

These tools aren't perfect and remain controversial for some software developers. While OpenAI has previously told Ars Technica that it uses Codex as a coding tool to help develop the Codex product itself, we also discovered, through hands-on experience, that these tools can be astonishingly fast at simple tasks but remain brittle beyond their training data and require human oversight for production work. The rough framework of a project tends to come fast and feels magical, but filling in the details involves tedious debugging and workarounds for limitations the agent cannot overcome on its own.

Bolin's post doesn't shy away from these engineering challenges. He discusses the inefficiency of quadratic prompt growth, performance issues caused by cache misses, and bugs the team discovered (like MCP tools being enumerated inconsistently) that they had to fix.

The level of technical detail is somewhat unusual for OpenAI, which has not published similar breakdowns of how other products like ChatGPT work internally, for example (there's a lot going on under that hood we'd like to know). But we've already seen how OpenAI treats Codex differently during our interview with them in December, noting that programming tasks seem ideally suited for large language models.

It's worth noting that both OpenAI and Anthropic open-source their coding CLI clients on GitHub, allowing developers to examine the implementation directly, whereas they don't do the same for ChatGPT or the Claude web interface.

An official look inside the loop

Bolin's post focuses on what he calls "the agent loop," which is the core logic that orchestrates interactions between the user, the AI model, and the software tools the model invokes to perform coding work.

As we wrote in December, at the center of every AI agent is a repeating cycle. The agent takes input from the user and prepares a textual prompt for the model. The model then generates a response, which either produces a final answer for the user or requests a tool call (such as running a shell command or reading a file). If the model requests a tool call, the agent executes it, appends the output to the original prompt, and queries the model again. This process repeats until the model stops requesting tools and instead produces an assistant message for the user.

That looping process has to start somewhere, and Bolin's post reveals how Codex constructs the initial prompt sent to OpenAI's Responses API, which handles model inference. The prompt is built from several components, each with an assigned role that determines its priority: system, developer, user, or assistant.

The instructions field comes from either a user-specified configuration file or base instructions bundled with the CLI. The tools field defines what functions the model can call, including shell commands, planning tools, web search capabilities, and any custom tools provided through Model Context Protocol (MCP) servers. The input field contains a series of items that describe the sandbox permissions, optional developer instructions, environment context like the current working directory, and finally the user's actual message.

As conversations continue, each new turn includes the complete history of previous messages and tool calls. This means the prompt grows with every interaction, which has performance implications. According to the post, because Codex does not use an optional "previous_response_id" parameter that would allow the API to reference stored conversation state, every request is fully stateless (that is, it sends the entire conversation history with each API call rather than the server retrieving it from memory). Bolin says this design choice simplifies things for API providers and makes it easier to support customers who opt into "Zero Data Retention," where OpenAI does not store user data.

The quadratic growth of prompts over a conversation is inefficient, but Bolin explains that prompt caching mitigates this issue somewhat. Cache hits only work for exact prefix matches within a prompt, which means Codex must carefully avoid operations that could cause cache misses. Changing the available tools, switching models, or modifying the sandbox configuration mid-conversation can all invalidate the cache and hurt performance.

The ever-growing prompt length is directly related to the context window, which limits how much text the AI model can process in a single inference call. Bolin writes that Codex automatically compacts conversations when token counts exceed a threshold, just as Claude Code does. Earlier versions of Codex required manual compaction via a slash command, but the current system uses a specialized API endpoint that compresses context while preserving summarized portions of the model's "understanding" of what happened through an encrypted content item.

Bolin says that future posts in his series will cover the CLI's architecture, tool implementation details, and Codex's sandboxing model.

Read full article

Comments

Yann LeCun helped create the technology behind today’s chatbots. Now he says many tech companies are on the wrong path to creating intelligent machines.

Concerns about the affordability of education, housing, health care, having a family and retirement are driving economic anxieties, a New York Times/Siena poll found.