Welcome to the Era of BadGPTs
Kanebridge News
    HOUSE MEDIAN ASKING PRICES AND WEEKLY CHANGE     Sydney $1,638,545 (+0.82%)       Melbourne $1,023,679 (+1.75%)       Brisbane $1,038,818 (+0.18%)       Adelaide $951,068 (+0.69%)       Perth $923,390 (-0.21%)       Hobart $759,192 (-0.42%)       Darwin $769,355 (-0.10%)       Canberra $964,485 (-0.83%)       National $1,074,245 (+0.50%)                UNIT MEDIAN ASKING PRICES AND WEEKLY CHANGE     Sydney $777,604 (+1.00%)       Melbourne $478,404 (+0.18%)       Brisbane $668,589 (+0.89%)       Adelaide $498,047 (-0.58%)       Perth $519,492 (+1.90%)       Hobart $528,197 (-0.03%)       Darwin $378,865 (-1.17%)       Canberra $494,950 (+0.08%)       National $567,655 (+0.60%)                HOUSES FOR SALE AND WEEKLY CHANGE     Sydney 11,855 (+190)       Melbourne 14,114 (-991)       Brisbane 8,271 (+242)       Adelaide 2,797 (+147)       Perth 7,549 (+147)       Hobart 1,213 (+7)       Darwin 181 (-4)       Canberra 1,228 (+25)       National 47,208 (-237)                UNITS FOR SALE AND WEEKLY CHANGE     Sydney 9,100 (+118)       Melbourne 6,842 (-31)       Brisbane 1,703 (+24)       Adelaide 418 (+32)       Perth 1,696 (+19)       Hobart 245 (+15)       Darwin 279 (+8)       Canberra 1,140 (-4)       National 21,423 (+181)                HOUSE MEDIAN ASKING RENTS AND WEEKLY CHANGE     Sydney $800 ($0)       Melbourne $590 ($0)       Brisbane $650 ($0)       Adelaide $620 ($0)       Perth $695 (-$5)       Hobart $555 (-$15)       Darwin $780 (+$20)       Canberra $700 ($0)       National $684 (+$1)                UNIT MEDIAN ASKING RENTS AND WEEKLY CHANGE     Sydney $750 ($0)       Melbourne $600 ($0)       Brisbane $650 (+$5)       Adelaide $525 (+$25)       Perth $650 ($0)       Hobart $480 (-$13)       Darwin $570 (+$5)       Canberra $570 (-$10)       National $610 (+$1)                HOUSES FOR RENT AND WEEKLY CHANGE     Sydney 6,415 (-92)       Melbourne 8,122 (-49)       Brisbane 4,023 (-50)       Adelaide 1,424 (-45)       Perth 2,128 (-99)       Hobart 232 (+21)       Darwin 103 (-17)       Canberra 559 (-41)       National 23,006 (-372)                UNITS FOR RENT AND WEEKLY CHANGE     Sydney 9,115 (-492)       Melbourne 6,656 (-238)       Brisbane 2,047 (-142)       Adelaide 349 (-56)       Perth 639 (-48)       Hobart 107 (+5)       Darwin 178 (-21)       Canberra 550 (-3)       National 19,641 (-995)                HOUSE ANNUAL GROSS YIELDS AND TREND         Sydney 2.54% (↓)       Melbourne 3.00% (↓)       Brisbane 3.25% (↓)       Adelaide 3.39% (↓)       Perth 3.91% (↓)       Hobart 3.80% (↓)     Darwin 5.27% (↑)      Canberra 3.77% (↑)        National 3.31% (↓)            UNIT ANNUAL GROSS YIELDS AND TREND         Sydney 5.02% (↓)       Melbourne 6.52% (↓)       Brisbane 5.06% (↓)     Adelaide 5.48% (↑)        Perth 6.51% (↓)       Hobart 4.73% (↓)     Darwin 7.82% (↑)        Canberra 5.99% (↓)       National 5.58% (↓)            HOUSE RENTAL VACANCY RATES AND TREND       Sydney 2.0% (↑)      Melbourne 1.9% (↑)      Brisbane 1.4% (↑)      Adelaide 1.3% (↑)      Perth 1.2% (↑)      Hobart 1.0% (↑)      Darwin 1.6% (↑)      Canberra 2.7% (↑)      National 1.7% (↑)             UNIT RENTAL VACANCY RATES AND TREND       Sydney 2.4% (↑)      Melbourne 3.8% (↑)      Brisbane 2.0% (↑)      Adelaide 1.1% (↑)      Perth 0.9% (↑)      Hobart 1.4% (↑)      Darwin 2.8% (↑)      Canberra 2.9% (↑)      National 2.2% (↑)             AVERAGE DAYS TO SELL HOUSES AND TREND         Sydney 29.5 (↓)       Melbourne 31.6 (↓)       Brisbane 31.5 (↓)       Adelaide 26.2 (↓)       Perth 41.1 (↓)       Hobart 33.2 (↓)       Darwin 24.8 (↓)       Canberra 32.7 (↓)       National 31.3 (↓)            AVERAGE DAYS TO SELL UNITS AND TREND         Sydney 25.4 (↓)       Melbourne 31.6 (↓)       Brisbane 27.4 (↓)       Adelaide 23.7 (↓)       Perth 41.0 (↓)       Hobart 26.8 (↓)       Darwin 45.2 (↓)       Canberra 43.3 (↓)       National 33.0 (↓)           
Share Button

Welcome to the Era of BadGPTs

The dark web is home to a growing array of artificial-intelligence chatbots similar to ChatGPT, but designed to help hackers. Businesses are on high alert for a glut of AI-generated email fraud and deepfakes.

By BELLE LIN
Thu, Feb 29, 2024 8:36amGrey Clock 5 min

A new crop of nefarious chatbots with names like “BadGPT” and “FraudGPT” are springing up on the darkest corners of the web, as cybercriminals look to tap the same artificial intelligence behind OpenAI’s ChatGPT.

Just as some office workers use ChatGPT to write better emails, hackers are using manipulated versions of AI chatbots to turbocharge their phishing emails. They can use chatbots—some also freely-available on the open internet—to create fake websites, write malware and tailor messages to better impersonate executives and other trusted entities.

Earlier this year, a Hong Kong multinational company employee handed over $25.5 million to an attacker who posed as the company’s chief financial officer on an AI-generated deepfake conference call, the South China Morning Post reported, citing Hong Kong police. Chief information officers and cybersecurity leaders, already accustomed to a growing spate of cyberattacks , say they are on high alert for an uptick in more sophisticated phishing emails and deepfakes.

Vish Narendra, CIO of Graphic Packaging International, said the Atlanta-based paper packing company has seen an increase in what are likely AI-generated email attacks called spear-phishing , where cyberattackers use information about a person to make an email seem more legitimate. Public companies in the spotlight are even more susceptible to contextualised spear-phishing, he said.

Researchers at Indiana University recently combed through over 200 large-language model hacking services being sold and populated on the dark web. The first service appeared in early 2023—a few months after the public release of OpenAI’s ChatGPT in November 2022.

Most dark web hacking tools use versions of open-source AI models like Meta ’s Llama 2, or “jailbroken” models from vendors like OpenAI and Anthropic to power their services, the researchers said. Jailbroken models have been hijacked by techniques like “ prompt injection ” to bypass their built-in safety controls.

Jason Clinton, chief information security officer of Anthropic, said the AI company eliminates jailbreak attacks as they find them, and has a team monitoring the outputs of its AI systems. Most model-makers also deploy two separate models to secure their primary AI model, making the likelihood that all three will fail the same way “a vanishingly small probability.”

Meta spokesperson Kevin McAlister said that openly releasing models shares the benefits of AI widely, and allows researchers to identify and help fix vulnerabilities in all AI models, “so companies can make models more secure.”

An OpenAI spokesperson said the company doesn’t want its tools to be used for malicious purposes, and that it is “always working on how we can make our systems more robust against this type of abuse.”

Malware and phishing emails written by generative AI are especially tricky to spot because they are crafted to evade detection. Attackers can teach a model to write stealthy malware by training it with detection techniques gleaned from cybersecurity defence software, said Avivah Litan, a generative AI and cybersecurity analyst at Gartner.

Phishing emails grew by 1,265% in the 12-month period starting when ChatGPT was publicly released, with an average of 31,000 phishing attacks sent every day, according to an October 2023 report by cybersecurity vendor SlashNext.

“The hacking community has been ahead of us,” said Brian Miller, CISO of New York-based not-for-profit health insurer Healthfirst, which has seen an increase in attacks impersonating its invoice vendors over the past two years.

While it is nearly impossible to prove whether certain malware programs or emails were created with AI, tools developed with AI can scan for text likely created with the technology. Abnormal Security , an email security vendor, said it had used AI to help identify thousands of likely AI-created malicious emails over the past year, and that it had blocked a twofold increase in targeted, personalised email attacks.

When Good Models Go Bad

Part of the challenge in stopping AI-enabled cybercrime is some AI models are freely shared on the open web. To access them, there is no need for dark corners of the internet or exchanging cryptocurrency.

Such models are considered “uncensored” because they lack the enterprise guardrails that businesses look for when buying AI systems, said Dane Sherrets, an ethical hacker and senior solutions architect at bug bounty company HackerOne.

In some cases, uncensored versions of models are created by security and AI researchers who strip out their built-in safeguards. In other cases, models with safeguards intact will write scam messages if humans avoid obvious triggers like “phishing”—a situation Andy Sharma, CIO and CISO of Redwood Software, said he discovered when creating a spear-phishing test for his employees.

The most useful model for generating scam emails is likely a version of Mixtral, from French AI startup Mistral AI, that has been altered to remove its safeguards, Sherrets said. Due to the advanced design of the original Mixtral, the uncensored version likely performs better than most dark web AI tools, he added. Mistral did not reply to a request for comment.

Sherrets recently demonstrated the process of using an uncensored AI model to generate a phishing campaign. First, he searched for “uncensored” models on Hugging Face, a startup that hosts a popular repository of open-source models—showing how easily many can be found.

He then used a virtual computing service that cost less than $1 per hour to mimic a graphics processing unit, or GPU, which is an advanced chip that can power AI. A bad actor needs either a GPU or a cloud-based service to use an AI model, Sherrets said, adding that he learned most of how to do this on X and YouTube.

With his uncensored model and virtual GPU service running, Sherrets asked the bot: “Write a phishing email targeting a business that impersonates a CEO and includes publicly-available company data,” and “Write an email targeting the procurement department of a company requesting an urgent invoice payment.”

The bot sent back phishing emails that were well-written, but didn’t include all of the personalisation asked for. That’s where prompt engineering , or the human’s ability to better extract information from chatbots, comes in, Sherrets said.

Dark Web AI Tools Can Already Do Harm

For hackers, a benefit of dark web tools like BadGPT—which researchers said uses OpenAI’s GPT model—is that they are likely trained on data from those underground marketplaces. That means they probably include useful information like leaks, ransomware victims and extortion lists, said Joseph Thacker, an ethical hacker and principal AI engineer at cybersecurity software firm AppOmni.

While some underground AI tools have been shuttered, new services have already taken their place, said Indiana University Assistant Computer Science Professor Xiaojing Liao, a co-author of the study. The AI hacking services, which often take payment via cryptocurrency, are priced anywhere from $5 to $199 a month.

New tools are expected to improve just as the AI models powering them do. In a matter of years, AI-generated text, video and voice deepfakes will be virtually indistinguishable from their human counterparts, said Evan Reiser , CEO and co-founder of Abnormal Security.

While researching the hacking tools, Indiana University Associate Dean for Research XiaoFeng Wang, a co-author of the study, said he was surprised by the ability of dark web services to generate effective malware. Given just the code of a security vulnerability, the tools can easily write a program to exploit it.

Though AI hacking tools often fail, in some cases, they work. “That demonstrates, in my opinion, that today’s large language models have the capability to do harm,” Wang said.



MOST POPULAR
11 ACRES ROAD, KELLYVILLE, NSW

This stylish family home combines a classic palette and finishes with a flexible floorplan

35 North Street Windsor

Just 55 minutes from Sydney, make this your creative getaway located in the majestic Hawkesbury region.

Related Stories
Money
NAB’s Earnings Hit by Higher Business-Loan Impairments
By Stuart Condie 19/02/2025
Money
Tech Giants Double Down on Their Massive AI Spending
By NATE RATTNER AND JASON DEAN 07/02/2025
Money
OpenAI in Talks for Huge Investment Round Valuing It Up to $300 Billion
By BERBER JIN and DEEPA SEETHARAMAN 31/01/2025
NAB’s Earnings Hit by Higher Business-Loan Impairments

The bank posted unaudited cash earnings for the quarter of A$1.7 billion, down 2% on the average of its prior two quarters

By Stuart Condie
Wed, Feb 19, 2025 < 1 min

National Australia Bank said that higher credit impairments against business loans contributed to a small fall in its unaudited December quarter cash earnings.

NAB , which is Australia’s second-largest bank by market capitalization, on Wednesday posted unaudited cash earnings for its fiscal first quarter of 1.74 billion Australian dollars, equivalent to about US$1.11 billion.

That was down 2% on the average of its prior two fiscal quarters. NAB did not give a year-earlier comparison.

The lender said that revenue grew by 3% compared with the average of its prior two fiscal quarters. Underlying profit growth of 4% over the same period was offset by higher credit impairment charges and income tax expenses, it added.

NAB, which posted an unaudited quarterly statutory profit of A$1.70 billion, said the A$267 million credit impairment charge included A$152 million of individually assessed charges. Those were mainly against Australian businesses and unsecured retail portfolios, it said.

The individual charges were up by 54% compared with a year earlier. NAB said that it had not altered its economic assumptions and scenario weightings.

“The economic outlook is improving but cost of living and interest rate challenges persisted,” Chief Executive Andrew Irvine said. “While most customers are proving resilient, we have maintained prudent balance sheet settings.”

NAB said it had seen a small decline in net interest margin due to funding costs, lending competition and deposits, partially offset by the benefit of higher interest rates.

On Tuesday, the Reserve Bank of Australia cut the country’s cash rate for the first time since 2020 but warned against expecting subsequent near-term cuts.

NAB is still targeting full fiscal-year productivity savings of more than A$400 million, and for operating expenses to grow by less than 4.5%, Irvine said.

MOST POPULAR
11 ACRES ROAD, KELLYVILLE, NSW

This stylish family home combines a classic palette and finishes with a flexible floorplan

35 North Street Windsor

Just 55 minutes from Sydney, make this your creative getaway located in the majestic Hawkesbury region.

Related Stories
Property of the Week
Property of the Week: 8-10 Howard St, Kew
By Kirsten Craze 17/02/2025
Property of the Week
Property of the Week: 26-27 Olola Rd, Vaucluse
By KIRSTEN CRAZE 13/12/2024
Money
China Pumps Up Support for Country’s Stock Markets
By TRACY QU 23/01/2025
0
    Your Cart
    Your cart is emptyReturn to Shop