April 17, 2025 • 15 mins
In this episode, we begin with an introduction and overview, leading into a discussion on the release and features of OpenAI's new models, o3 and o4-mini. We assess their benchmark performance and efficiency, and explore their real-world applications, particularly in coding. The episode highlights the announcement of Codex CLI and its availability to users, followed by a look at human-AI collaboration and potential acquisitions. We address industry competition, antitrust issues, and concerns about model access, along with intellectual property challenges faced by OpenAI. The conversation also covers the Stargate project and its implications for OpenAI's expansion in the UK. The episode concludes with closing remarks and sign-off.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Are we on the brink of a new era in artificialintelligence, where AI systems think, reason,
(00:05):
and act like never before!?
Welcome to The OpenAI Daily Brief, your go-tofor the latest AI updates.
Today is Thursday, April 17th, 2025.
Here’s what you need to know about OpenAI'sgame-changing models, o3 and o4-mini.
Let’s dive in.
(00:26):
In what feels like a significant evolution ofAI capabilities, OpenAI has released two new
models that push the boundaries of what we'vecome to expect from large language models.
The company's latest offerings—o3 ando4-mini—are being touted as their most capable
releases to date.
Mark Chen, OpenAI's research lead, describedthis launch as "a qualitative step into the
(00:50):
future" during the announcement livestream.
These models are not just about rawintelligence; they're about how they integrate
with other tools.
For the first time, OpenAI's reasoning modelscan independently decide when and how to use
the full suite of ChatGPT tools, like webbrowsing, Python code execution, image
(01:11):
analysis, and image generation, to solvecomplex problems.
Greg Brockman highlighted during the launchthat "these aren't just models.
They're really AI systems." This subtledistinction marks a shift in how these models
operate.
One of the standout features is how thesemodels handle visual information.
(01:32):
Unlike previous iterations that could merely"see" images, o3 and o4-mini can "think with
images." They manipulate them in theirreasoning process—rotating, zooming, or
transforming uploaded photos as part ofproblem-solving.
This capability has led to state-of-the-artperformance on multimodal benchmarks,
(01:52):
interpreting even blurry or oddly orientedvisuals with impressive accuracy.
The benchmark results are indeed noteworthy.
On AIME 2024, a challenging math competition,o4-mini achieved 93.4% accuracy without tools.
(02:13):
On Codeforces, a competitive programmingplatform, o3 with terminal access reached an
ELO of 2706—placing it among the topcompetitors globally.
And on PhD-level science questions in the GPQADiamond benchmark, o3 without tools scored
83.3% accuracy, outperforming previous modelsby a significant margin.
(02:39):
What's particularly remarkable about o4-mini isits efficiency.
Despite being smaller and more cost-efficientthan o3, it achieves comparable or even
superior performance on certain benchmarks.
This efficiency translates to higher usagelimits, making it particularly valuable for
high-volume applications.
(03:01):
Behind these improvements lies a massiveincrease in training compute compared to
previous models.
OpenAI reports investing more than 10 times thetraining compute of o1 to produce o3.
This investment validates the company's beliefthat "more compute equals better performance,"
a principle that holds true for reinforcementlearning just as it did for pre-training in the
(03:25):
GPT series.
Perhaps the most intriguing demonstrations camefrom researchers showing how these models solve
real-world problems.
In one example, a researcher uploaded a blurryphysics poster from a decade-old internship and
asked the model to extract key information andcompare it with recent literature.
(03:46):
The model not only navigated the complex posterbut also identified that the specific result
wasn't included, calculated it based onavailable data, and compared it with current
research—a task that would have taken days forthe researcher.
The models also showed impressive codingcapabilities.
During the demonstration, o3 successfullydebugged a complex issue in a Python
(04:10):
mathematical package by navigating throughsource code, identifying inheritance problems,
and applying the correct fix—all while runningappropriate tests to verify the solution
worked.
Alongside these releases, OpenAI announced anew developer tool called Codex CLI, a
command-line interface that connects the modelsdirectly to users' computers.
(04:32):
This tool allows models to run commands locallywith safety constraints, potentially reshaping
how developers interact with their machines.
To support this vision, OpenAI is launching a$1 million initiative to provide API credits
for open-source projects using these newmodels.
For users, these models are already replacingprevious versions in ChatGPT.
(04:56):
Plus, Pro, and Team users now have access too3, o4-mini, and o4-mini-high in the model
selector, with Enterprise and Edu users gainingaccess next week.
The models are also available to developersthrough the Chat Completions API and Responses
API.
As impressive as these advancements are, theyraise important questions about the evolving
(05:19):
relationship between humans and increasinglycapable AI systems.
With models now able to chain hundreds of toolcalls together and independently navigate
complex decision trees, we're seeing the firstglimpses of truly agentic AI—systems that can
independently execute multi-step tasks on auser's behalf.
(05:41):
What's clear is that the boundary between"using a model" and "working with an AI system"
continues to blur.
As these systems become more capable ofindependent problem-solving, the relationship
becomes increasingly collaborative rather thaninstructional—a shift that may fundamentally
change how we think about AI tools in the yearsahead.
(06:03):
OpenAI is making headlines once again, but thistime it's not just about new models.
They're reportedly in advanced talks to acquireWindsurf, a coding assistant company, for over
three billion dollars.
Now, that's a hefty price tag, and if this dealgoes through, it would mark OpenAI's largest
acquisition to date.
(06:24):
So why is this a big deal?
Let's unpack it.
Picture this (06:28):
OpenAI, known for its cutting-edge AI models like ChatGPT and DALL-E,
is now looking to expand its footprint in therealm of AI applications.
They've done amazing in-house work, butacquiring a company like Windsurf could
accelerate their capabilities in codingassistance.
(06:48):
Imagine if Microsoft had only remained anoperating system company—OpenAI isn't about to
make that mistake.
They're eyeing growth, and this acquisitioncould be a strategic move to stay ahead in the
AI race.
But hold on, it's not a done deal just yet.
While talks are advanced, things could gettricky due to a recent leak reported by
(07:09):
Bloomberg.
Plus, most of that three billion dollars mightactually be in OpenAI stock.
Windsurf, formerly known as Codeium, has beenconsidering new venture capital funding too.
They've already raised two hundred fortymillion dollars, with their last valuation at
one point two five billion dollars.
(07:29):
So, there's a lot at play here.
Now, here's where it gets even more intriguing.
Windsurf isn't the only player in town.
Anysphere, a rival to Windsurf, raised seedfunding back in 2023 from the OpenAI Startup
Fund.
And they're reportedly raising new funding at aten billion dollar valuation.
(07:49):
It's like a chess game where every move counts,and OpenAI's next move could set off a domino
effect in the industry.
Let's zoom out for a moment.
If this acquisition goes through, it could posesome interesting questions for antitrust
regulators.
The Federal Trade Commission is already in alegal battle with Meta over acquisitions it
(08:10):
previously approved.
Could they see OpenAI's potential acquisitionof Windsurf as a similar situation?
Or might they view it as fostering competition,especially against Microsoft Copilot?
The dynamic between OpenAI and Microsoft iscomplex, almost like frenemies.
It'll be fascinating to see how this plays out.
(08:33):
OpenAI's latest move is causing quite a stir inthe AI community.
They're tightening access to their advancedmodels, and it's not just about curbing misuse.
The real story here is about protecting theirintellectual property.
OpenAI is now requiring developers to verifytheir identities with government IDs before
they get access to these powerful tools.
(08:54):
It's a bold move, but what's driving it?
Well, imagine this (08:57):
You've built something groundbreaking, but soon you find out that
others are copying your work without yourconsent.
That's kind of what's happening here.
A research paper from Copyleaks, a company thatspecializes in AI content detection, has
highlighted that a whopping seventy-fourpercent of outputs from a rival model,
(09:18):
DeepSeek-R1, were classified as being writtenby OpenAI.
This isn't just overlap—it's outright mimicry.
Now, Copyleaks has developed a system thatidentifies the "fingerprints" of major AI
models.
These are like linguistic signatures thatpersist across tasks, topics, and prompts,
(09:38):
making it possible to trace back to the source.
So, when DeepSeek's outputs show a seventy-fourpercent similarity to OpenAI's models, it
raises some serious eyebrows.
OpenAI hasn't officially commented on this, butthey've hinted at concerns that DeepSeek might
have "inappropriately distilled" their models.
(09:58):
Distillation is a process where new models aretrained using the outputs of existing ones.
It's a common technique, but doing it withoutpermission could violate OpenAI's terms of
service.
Critics are quick to point out that OpenAI'searly models were trained using web-scraped
data, often without explicit consent from thecreators.
(10:21):
So, is it hypocritical for OpenAI to nowcomplain about others using their outputs
similarly?
Alon Yamin, CEO of Copyleaks, argues that theissue really boils down to consent and
transparency.
Training on copyrighted human content withoutpermission is one thing, but using proprietary
AI outputs is more like reverse-engineeringsomeone else's product.
(10:45):
As AI companies race to build more advancedmodels, the debate over who owns what—and who
can train on whom—is heating up.
Tools like Copyleaks' digital fingerprintingsystem might offer a way to trace and verify
authorship at the model level.
For OpenAI and its rivals, this could be both ablessing and a warning.
(11:06):
OpenAI's growth has been nothing short ofmind-boggling.
Just a few years ago, they were primarily knownfor their GPT models, but now, they've expanded
their reach across various sectors, becoming ahousehold name in artificial intelligence.
This rapid expansion, however, isn't withoutits challenges.
Picture this (11:27):
you're driving a car at breakneck speed on a winding road.
It’s exhilarating, but you’ve got to keep aneye on the roadblocks ahead.
For OpenAI, their explosive growth is like thathigh-speed drive.
They’ve achieved incredible success, butthey're also facing some significant hurdles.
First and foremost, there's the issue ofscaling.
(11:48):
As OpenAI's operations expand, they're dealingwith the growing pains that come with scaling
up their infrastructure and workforce.
The demand for AI services is skyrocketing, andmeeting that demand requires a robust system
that can handle the load without buckling underpressure.
Then there's the competition.
The AI landscape is more crowded than ever,with tech giants and startups alike vying for a
(12:13):
piece of the pie.
OpenAI might be leading the pack now, butstaying ahead requires constant innovation and
adaptation, especially when competitors arenipping at their heels.
And let's not forget the regulatory landscape.
As AI technologies become more ingrained in ourdaily lives, the call for regulation grows
louder.
(12:34):
OpenAI must navigate a complex web of globalregulations, ensuring compliance while also
advocating for responsible AI development.
Despite these challenges, OpenAI's growth storyis a testament to their innovation and vision.
They've managed to capture the imagination ofthe world, and their journey is far from over.
(12:55):
The road ahead is filled with bothopportunities and obstacles, and it'll be
fascinating to see how they maneuver throughthis dynamic landscape.
OpenAI's latest venture, the massive $500billion Stargate project, is making waves as it
considers expanding its AI infrastructureoverseas, with the United Kingdom emerging as a
(13:15):
potential candidate.
This bold move is backed by heavyweights likeSoftBank and Oracle, and it's all about
establishing a robust data-center network tosupport AI development on a global scale.
Now, why the UK?
Well, it's all about the pro-innovation stancethat British Prime Minister Keir Starmer is
championing.
(13:37):
He's promising to streamline regulations, makepublic data more accessible to researchers, and
create dedicated zones for data centers—effortsthat have certainly caught the eye of
Stargate's stakeholders.
But the UK is not the only player in this game.
Germany and France are also on the radar,offering attractive conditions for such massive
(13:58):
infrastructure projects.
It's a competitive landscape, but the UK’scommitment to becoming an AI "superpower" might
just tip the scales in their favor.
This isn't just about building data centers;it's about positioning the UK as a leader in AI
innovation.
With OpenAI CEO Sam Altman expressing a keeninterest in bringing Stargate-like programs to
(14:22):
Europe, this could be a significant step inestablishing the continent as a powerhouse in
AI.
The Stargate project was unveiled by U.S.
President Donald Trump as a private sectorinitiative to bolster AI infrastructure, aiming
to outpace global rivals.
It's a testament to the growing investorenthusiasm in AI, driven by the widespread
(14:45):
adoption of chatbots and sophisticated AIagents.
So, what does this mean for the future?
If Stargate decides to invest in the UK, itcould accelerate the nation's ascent in the AI
industry, opening up new opportunities andfostering innovation.
It’s an exciting prospect that could reshapethe global AI landscape.
(15:08):
That’s it for today’s OpenAI Weekly Brief.
OpenAI's potential investment in the UK's AIinfrastructure through the Stargate project
highlights the growing importance ofinternational collaboration in AI development.
Thanks for tuning in—subscribe to stay updated.
This is Michelle, signing off.
Until next time.
Are we on the brink of a new era in artificialintelligence, where AI systems think, reason,
(00:05):
and act like never before!?
Welcome to The OpenAI Daily Brief, your go-tofor the latest AI updates.
Today is Thursday, April 17th, 2025.
Here’s what you need to know about OpenAI'sgame-changing models, o3 and o4-mini.
Let’s dive in.
(00:26):
In what feels like a significant evolution ofAI capabilities, OpenAI has released two new
models that push the boundaries of what we'vecome to expect from large language models.
The company's latest offerings—o3 ando4-mini—are being touted as their most capable
releases to date.
Mark Chen, OpenAI's research lead, describedthis launch as "a qualitative step into the
(00:50):
future" during the announcement livestream.
These models are not just about rawintelligence; they're about how they integrate
with other tools.
For the first time, OpenAI's reasoning modelscan independently decide when and how to use
the full suite of ChatGPT tools, like webbrowsing, Python code execution, image
(01:11):
analysis, and image generation, to solvecomplex problems.
Greg Brockman highlighted during the launchthat "these aren't just models.
They're really AI systems." This subtledistinction marks a shift in how these models
operate.
One of the standout features is how thesemodels handle visual information.
(01:32):
Unlike previous iterations that could merely"see" images, o3 and o4-mini can "think with
images." They manipulate them in theirreasoning process—rotating, zooming, or
transforming uploaded photos as part ofproblem-solving.
This capability has led to state-of-the-artperformance on multimodal benchmarks,
(01:52):
interpreting even blurry or oddly orientedvisuals with impressive accuracy.
The benchmark results are indeed noteworthy.
On AIME 2024, a challenging math competition,o4-mini achieved 93.4% accuracy without tools.
(02:13):
On Codeforces, a competitive programmingplatform, o3 with terminal access reached an
ELO of 2706—placing it among the topcompetitors globally.
And on PhD-level science questions in the GPQADiamond benchmark, o3 without tools scored
83.3% accuracy, outperforming previous modelsby a significant margin.
(02:39):
What's particularly remarkable about o4-mini isits efficiency.
Despite being smaller and more cost-efficientthan o3, it achieves comparable or even
superior performance on certain benchmarks.
This efficiency translates to higher usagelimits, making it particularly valuable for
high-volume applications.
(03:01):
Behind these improvements lies a massiveincrease in training compute compared to
previous models.
OpenAI reports investing more than 10 times thetraining compute of o1 to produce o3.
This investment validates the company's beliefthat "more compute equals better performance,"
a principle that holds true for reinforcementlearning just as it did for pre-training in the
(03:25):
GPT series.
Perhaps the most intriguing demonstrations camefrom researchers showing how these models solve
real-world problems.
In one example, a researcher uploaded a blurryphysics poster from a decade-old internship and
asked the model to extract key information andcompare it with recent literature.
(03:46):
The model not only navigated the complex posterbut also identified that the specific result
wasn't included, calculated it based onavailable data, and compared it with current
research—a task that would have taken days forthe researcher.
The models also showed impressive codingcapabilities.
During the demonstration, o3 successfullydebugged a complex issue in a Python
(04:10):
mathematical package by navigating throughsource code, identifying inheritance problems,
and applying the correct fix—all while runningappropriate tests to verify the solution
worked.
Alongside these releases, OpenAI announced anew developer tool called Codex CLI, a
command-line interface that connects the modelsdirectly to users' computers.
(04:32):
This tool allows models to run commands locallywith safety constraints, potentially reshaping
how developers interact with their machines.
To support this vision, OpenAI is launching a$1 million initiative to provide API credits
for open-source projects using these newmodels.
For users, these models are already replacingprevious versions in ChatGPT.
(04:56):
Plus, Pro, and Team users now have access too3, o4-mini, and o4-mini-high in the model
selector, with Enterprise and Edu users gainingaccess next week.
The models are also available to developersthrough the Chat Completions API and Responses
API.
As impressive as these advancements are, theyraise important questions about the evolving
(05:19):
relationship between humans and increasinglycapable AI systems.
With models now able to chain hundreds of toolcalls together and independently navigate
complex decision trees, we're seeing the firstglimpses of truly agentic AI—systems that can
independently execute multi-step tasks on auser's behalf.
(05:41):
What's clear is that the boundary between"using a model" and "working with an AI system"
continues to blur.
As these systems become more capable ofindependent problem-solving, the relationship
becomes increasingly collaborative rather thaninstructional—a shift that may fundamentally
change how we think about AI tools in the yearsahead.
(06:03):
OpenAI is making headlines once again, but thistime it's not just about new models.
They're reportedly in advanced talks to acquireWindsurf, a coding assistant company, for over
three billion dollars.
Now, that's a hefty price tag, and if this dealgoes through, it would mark OpenAI's largest
acquisition to date.
(06:24):
So why is this a big deal?
Let's unpack it.
Picture this (06:28):
OpenAI, known for its cutting-edge AI models like ChatGPT and DALL-E,
is now looking to expand its footprint in therealm of AI applications.
They've done amazing in-house work, butacquiring a company like Windsurf could
accelerate their capabilities in codingassistance.
(06:48):
Imagine if Microsoft had only remained anoperating system company—OpenAI isn't about to
make that mistake.
They're eyeing growth, and this acquisitioncould be a strategic move to stay ahead in the
AI race.
But hold on, it's not a done deal just yet.
While talks are advanced, things could gettricky due to a recent leak reported by
(07:09):
Bloomberg.
Plus, most of that three billion dollars mightactually be in OpenAI stock.
Windsurf, formerly known as Codeium, has beenconsidering new venture capital funding too.
They've already raised two hundred fortymillion dollars, with their last valuation at
one point two five billion dollars.
(07:29):
So, there's a lot at play here.
Now, here's where it gets even more intriguing.
Windsurf isn't the only player in town.
Anysphere, a rival to Windsurf, raised seedfunding back in 2023 from the OpenAI Startup
Fund.
And they're reportedly raising new funding at aten billion dollar valuation.
(07:49):
It's like a chess game where every move counts,and OpenAI's next move could set off a domino
effect in the industry.
Let's zoom out for a moment.
If this acquisition goes through, it could posesome interesting questions for antitrust
regulators.
The Federal Trade Commission is already in alegal battle with Meta over acquisitions it
(08:10):
previously approved.
Could they see OpenAI's potential acquisitionof Windsurf as a similar situation?
Or might they view it as fostering competition,especially against Microsoft Copilot?
The dynamic between OpenAI and Microsoft iscomplex, almost like frenemies.
It'll be fascinating to see how this plays out.
(08:33):
OpenAI's latest move is causing quite a stir inthe AI community.
They're tightening access to their advancedmodels, and it's not just about curbing misuse.
The real story here is about protecting theirintellectual property.
OpenAI is now requiring developers to verifytheir identities with government IDs before
they get access to these powerful tools.
(08:54):
It's a bold move, but what's driving it?
Well, imagine this (08:57):
You've built something groundbreaking, but soon you find out that
others are copying your work without yourconsent.
That's kind of what's happening here.
A research paper from Copyleaks, a company thatspecializes in AI content detection, has
highlighted that a whopping seventy-fourpercent of outputs from a rival model,
(09:18):
DeepSeek-R1, were classified as being writtenby OpenAI.
This isn't just overlap—it's outright mimicry.
Now, Copyleaks has developed a system thatidentifies the "fingerprints" of major AI
models.
These are like linguistic signatures thatpersist across tasks, topics, and prompts,
(09:38):
making it possible to trace back to the source.
So, when DeepSeek's outputs show a seventy-fourpercent similarity to OpenAI's models, it
raises some serious eyebrows.
OpenAI hasn't officially commented on this, butthey've hinted at concerns that DeepSeek might
have "inappropriately distilled" their models.
(09:58):
Distillation is a process where new models aretrained using the outputs of existing ones.
It's a common technique, but doing it withoutpermission could violate OpenAI's terms of
service.
Critics are quick to point out that OpenAI'searly models were trained using web-scraped
data, often without explicit consent from thecreators.
(10:21):
So, is it hypocritical for OpenAI to nowcomplain about others using their outputs
similarly?
Alon Yamin, CEO of Copyleaks, argues that theissue really boils down to consent and
transparency.
Training on copyrighted human content withoutpermission is one thing, but using proprietary
AI outputs is more like reverse-engineeringsomeone else's product.
(10:45):
As AI companies race to build more advancedmodels, the debate over who owns what—and who
can train on whom—is heating up.
Tools like Copyleaks' digital fingerprintingsystem might offer a way to trace and verify
authorship at the model level.
For OpenAI and its rivals, this could be both ablessing and a warning.
(11:06):
OpenAI's growth has been nothing short ofmind-boggling.
Just a few years ago, they were primarily knownfor their GPT models, but now, they've expanded
their reach across various sectors, becoming ahousehold name in artificial intelligence.
This rapid expansion, however, isn't withoutits challenges.
Picture this (11:27):
you're driving a car at breakneck speed on a winding road.
It’s exhilarating, but you’ve got to keep aneye on the roadblocks ahead.
For OpenAI, their explosive growth is like thathigh-speed drive.
They’ve achieved incredible success, butthey're also facing some significant hurdles.
First and foremost, there's the issue ofscaling.
(11:48):
As OpenAI's operations expand, they're dealingwith the growing pains that come with scaling
up their infrastructure and workforce.
The demand for AI services is skyrocketing, andmeeting that demand requires a robust system
that can handle the load without buckling underpressure.
Then there's the competition.
The AI landscape is more crowded than ever,with tech giants and startups alike vying for a
(12:13):
piece of the pie.
OpenAI might be leading the pack now, butstaying ahead requires constant innovation and
adaptation, especially when competitors arenipping at their heels.
And let's not forget the regulatory landscape.
As AI technologies become more ingrained in ourdaily lives, the call for regulation grows
louder.
(12:34):
OpenAI must navigate a complex web of globalregulations, ensuring compliance while also
advocating for responsible AI development.
Despite these challenges, OpenAI's growth storyis a testament to their innovation and vision.
They've managed to capture the imagination ofthe world, and their journey is far from over.
(12:55):
The road ahead is filled with bothopportunities and obstacles, and it'll be
fascinating to see how they maneuver throughthis dynamic landscape.
OpenAI's latest venture, the massive $500billion Stargate project, is making waves as it
considers expanding its AI infrastructureoverseas, with the United Kingdom emerging as a
(13:15):
potential candidate.
This bold move is backed by heavyweights likeSoftBank and Oracle, and it's all about
establishing a robust data-center network tosupport AI development on a global scale.
Now, why the UK?
Well, it's all about the pro-innovation stancethat British Prime Minister Keir Starmer is
championing.
(13:37):
He's promising to streamline regulations, makepublic data more accessible to researchers, and
create dedicated zones for data centers—effortsthat have certainly caught the eye of
Stargate's stakeholders.
But the UK is not the only player in this game.
Germany and France are also on the radar,offering attractive conditions for such massive
(13:58):
infrastructure projects.
It's a competitive landscape, but the UK’scommitment to becoming an AI "superpower" might
just tip the scales in their favor.
This isn't just about building data centers;it's about positioning the UK as a leader in AI
innovation.
With OpenAI CEO Sam Altman expressing a keeninterest in bringing Stargate-like programs to
(14:22):
Europe, this could be a significant step inestablishing the continent as a powerhouse in
AI.
The Stargate project was unveiled by U.S.
President Donald Trump as a private sectorinitiative to bolster AI infrastructure, aiming
to outpace global rivals.
It's a testament to the growing investorenthusiasm in AI, driven by the widespread
(14:45):
adoption of chatbots and sophisticated AIagents.
So, what does this mean for the future?
If Stargate decides to invest in the UK, itcould accelerate the nation's ascent in the AI
industry, opening up new opportunities andfostering innovation.
It’s an exciting prospect that could reshapethe global AI landscape.
(15:08):
That’s it for today’s OpenAI Weekly Brief.
OpenAI's potential investment in the UK's AIinfrastructure through the Stargate project
highlights the growing importance ofinternational collaboration in AI development.
Thanks for tuning in—subscribe to stay updated.
This is Michelle, signing off.
Until next time.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.